编程练习——huffman编码

这个类的名字叫huffman Tree，但是不仅仅是生成huffman Tree，因为Huffman树的生成要使用优先队列，也就是堆。stl中其实有这个的实现。但是我机器里的vs2005这个堆的使用有点小bug（估计是stl中使用的就是数组存储，而且数组一旦固定大小，就再也无法改变，其实这样实现失去了堆的某些特性。所以我还是自己实现了一下。顺便练习下。）。刚开始的时候就采用int作为编码的存储单元。后来发现huffman是可变长编码。所以使用int作为编码单元就失去了huffman的特性。于是寻找bit数组。找到了bitset这个stl类，却发现这个是不可变的。无语。只得自己实现一个可变大小的bit vector。然后近乎重写了这个Huffman的编码和解码。不过也不是很难。在这之间又陆陆续续的发现了一系列的bug和编译错误。对以后的编程大有启发。

1. /*createdbychicochen
2. *date2008/10/25
3. */
6. #ifndef_HUFFMAN_TREE_
7. #define_HUFFMAN_TREE_
8. #include"Heap.h"
9. #include"TreeNode.h"
10. #include"BinaryTree.h"
11. #include<map>
12. #include"../../ArithmeticTest/ArithmeticTest/BITVector.h"
13. usingnamespacestd;
15. //thisclassisusedforthecombinfunctioninhuffmanTreeclass.
16. classIntCombin
17. {
18. public:
19. staticintAdd(inti,intj)
20. {
21. returni+j;
22. }
23. };
25. //thisclassforcomparetreenode
26. template<classT>
27. classTreeNodeCompare
28. {
29. public:
30. staticboolUp(TreeNode<T>*d1,TreeNode<T>*d2)
31. {
32. returnLt(d1->getData(),d2->getData());
33. }
35. staticboolEq(T&d1,T&d2)
36. {
37. if(d1==d2)
38. returntrue;
39. returnfalse;
40. }
42. staticboolGt(T&d1,T&d2)
43. {
44. if(d1>d2)
45. returntrue;
46. returnfalse;
47. }
49. staticboolLt(T&d1,T&d2)
50. {
51. if(d1<d2)
52. returntrue;
53. returnfalse;
54. }
55. };
57. //thisclassisforstoringhuffmancode.
58. //bitsandthelengthofbits
59. classHuffmanCode
60. {
61. public:
62. BITVectorbits;
63. intlen;
64. HuffmanCode()
65. {
66. len=0;
67. }
69. HuffmanCode(constHuffmanCode&codes):bits(codes.bits)
70. {
71. len=codes.len;
73. }
75. constHuffmanCode&operator=(constHuffmanCode&codes)
76. {
77. if(this!=&codes)
78. {
79. this->len=codes.len;
80. this->bits=codes.bits;
81. }
82. return*this;
83. }
85. ~HuffmanCode()
86. {
88. }
89. };
91. template<classT,classCmp,classCombin>
92. classHuffmanTree:publicBinaryTree<T>
93. {
94. private:
95. //storethedata,andsortit,thenforbuildingahuffmantree
96. Heap<TreeNode<T>*,Cmp>huffmanQueue;
98. constunsignedintmask;
99. //themaxnumbercanbeshifted,ifitisunsignedintthenitwillbe32.
100. constintmaxShiftNum;
103. TreeNode<T>*initTree()
104. //initthehuffmantree
105. {
106. TreeNode<T>*combinNode=NULL;
107. TtempData;
108. while(!this->huffmanQueue.IsEmpty())
109. {
110. //fetchtwosmalldataandgeneratealargedata
111. TreeNode<T>*node1=this->huffmanQueue.Top();
112. combinNode=node1;
113. this->huffmanQueue.RemoveTop();
114. if(!this->huffmanQueue.IsEmpty())
115. {
116. TreeNode<T>*node2=this->huffmanQueue.Top();
117. this->huffmanQueue.RemoveTop();
118. tempData=Combin::Add(node1->getData(),node2->getData());
121. if(Cmp::Lt(node1->getData(),node2->getData()))
122. {
123. //hereiscomparingthedata1ofnode1andthedata2ofnode2
124. //Cmp:Ltmeans'<'andCmp:Gtmeans'>'
126. combinNode=newTreeNode<T>(tempData,node1,node2);
127. }
128. else
129. {
130. combinNode=newTreeNode<T>(tempData,node2,node1);
131. }
132. this->huffmanQueue.Insert(combinNode);
133. }
134. else
135. {
136. break;
137. }
139. }
140. returncombinNode;
141. }
145. //thefinalhuffmancode
146. map<T,HuffmanCode>huffmanTable;
149. voidSelfEncode(TreeNode<T>*subRoot,unsignedintbitcode,intnum,map<T,HuffmanCode>&ht)
150. //encodingthedatafromahuffmantree.
151. //Theleftbranchis0,andtherightbranchis1
152. {
154. if(subRoot!=NULL)
155. {
157. if(subRoot->left==NULL&&subRoot->right==NULL)
158. {
159. HuffmanCodehc;
160. hc.bits.SetInt(bitcode,0,maxShiftNum-num);
161. hc.len=maxShiftNum-num;
162. ht[subRoot->getData()]=hc;
163. return;
164. }
165. else
166. {
167. if(num<=0)
168. {
169. throw"thehuffmanistoodeep!";
170. }
173. SelfEncode(subRoot->left,bitcode,num-1,ht);
174. bitcode=bitcode|(0x80000000>>(maxShiftNum-num));
175. SelfEncode(subRoot->right,bitcode,num-1,ht);
176. }
177. }
180. }
183. voidDecodeFromTree(TreeNode<T>*subRoot,BITVector&bits,intindex,constint&len,vector<T>&decode)
184. //decodingthecode.usethecodetosearchthedatafromthehuffmanTree.
185. //hereyoucancreateyouownmap<code,data>deHuffmanTable.
186. //butIcannotanalyzethespacethetablewilluse,
187. //soIfindthedatabysearchinghuffmantree
188. {
189. if(index==len)
190. {
191. if(subRoot->getLeft()==NULL&&subRoot->getRight()==NULL)
192. {
194. decode.push_back(subRoot->getData());
195. return;
196. }
197. throw"codelengthisnotmatchthebitslength.HuffmanTreeconstructerror.";
198. }
199. if(subRoot==NULL)
200. {
201. throw"decodeerror!";
203. }
204. elseif(subRoot->getLeft()==NULL&&subRoot->getRight()==NULL)
205. {
207. decode.push_back(subRoot->getData());
208. returnDecodeFromTree(this->root,bits,index,len,decode);
209. }
211. if(bits.Get(index)==0)
212. {
213. returnDecodeFromTree(subRoot->getLeft(),bits,index+1,len,decode);
214. }
215. elseif(bits.Get(index)==1)
216. {
218. returnDecodeFromTree(subRoot->getRight(),bits,index+1,len,decode);
219. }
220. else
221. {
222. throw"codeerror!";
223. }
226. }
229. unsignedintFindCodeFromTable(T&data)
230. //findthedatacodefromhuffmantable
231. //maybenotefficient.
232. {
233. returnthis->huffmanTable[data];
234. }
236. public:
238. HuffmanTree(Tt[],intn):
239. BinaryTree<T>(),mask(0x80000000),maxShiftNum(sizeof(unsignedint)*8)
240. //thearrayttypeisT,andthenumberisn
241. {
242. if(n==0)
243. return;
244. TreeNode<T>*node;
245. for(inti=0;i<n;i++)
246. {
247. node=newTreeNode<T>(t[i]);
248. this->huffmanQueue.Insert(node);
249. }
250. this->root=initTree();
251. }
253. ~HuffmanTree()
254. {
255. //destroy
257. }
260. voidSelfEncode()
261. //convertthehuffmanTreeintohuffmanTable
262. //unsignedintcodeis32bits,sothehuffmantreehasonlylessthan33layer.
263. {
264. strings="";
266. SelfEncode(this->root,0,maxShiftNum,this->huffmanTable);
268. }
272. voidDecode(HuffmanCode&huffmanCode,vector<T>&decode)
273. //usebittofindthedatanode.
274. {
276. returnDecodeFromTree(this->root,huffmanCode.bits,0,huffmanCode.len,decode);
278. }
280. HuffmanCodeEncode(Tinfo[],intn)
281. //nissize
282. {
283. HuffmanCodehc;
285. for(inti=0;i<n;i++)
286. {
288. hc.bits.SetBitVector(((HuffmanCode)this->huffmanTable[info[i]]).bits,hc.len,((HuffmanCode)this->huffmanTable[info[i]]).len);
289. hc.len+=((HuffmanCode)this->huffmanTable[info[i]]).len;
290. }
292. returnhc;
293. }
296. voidPrintHuffmanTable()
297. //printthehuffmantable
298. //printthepairdata<->code
299. {
300. intlen=this->huffmanTable.size();
301. cout<<"i/tdata/tcode/n";
302. intcount=0;
303. map<T,HuffmanCode>::iteratori=this->huffmanTable.begin();
304. for(;i!=this->huffmanTable.end();i++)
305. {
306. cout<<count++<<"/t"<<(*i).first<<"/t";
307. (*i).second.bits.PrintfZeroOne(0,(*i).second.len);
308. cout<<endl;
309. }
311. }
313. };
316. #endif

这个类设计方面还有些不足之处。比如encode方法

测试一下：

1. inta[10]={12,224,33,32,1,91,35,34,36,291};
2. HuffmanTree<int,TreeNodeCompare<int>,IntCombin>ht(a,10);
3. ht.SelfEncode();
4. ht.PrintHuffmanTable();
5. ht.printTree();
6. cout<<endl;
7. intinfo[]={33,34,33};
8. HuffmanCodehc=ht.Encode(info,3);
9. hc.bits.PrintfZeroOne(0,hc.len);
10. vector<int>code;
11. ht.Decode(hc,code);
12. for(inti=0;i!=code.size();i++)
13. {
15. cout<<code[i]<<endl;
16. }
18. return0;