textquery.cpp

文件压缩与解压

# include "TextQuery.h"

void TextQuery::BuildMap(std::ifstream& in, std::string file_name)
{
	m.clear();
	strall.clear();
	huffumancode.clear();
	str="";
	sum=0;
	in.open(file_name.c_str());
	if(!in)
	{
		cerr<<"The file does not exsit!"<<endl;
		return;
	}
	string strread;
	while(getline(in,strread))
	{
		strall+=strread;
		if(!in.eof())
			strall+='\n';
	}
	for(size_t i=0;i<strall.size();i++)
	{
		m[strall[i]]++;
	}
	tree=HuffumanTree();
	in.close();
	in.clear();
}

ExtBinTree<char>* TextQuery::HuffumanTree()
{
	ExtBinTree<char>* newtree=NULL;
	ExtBinTree<char> first,second;
	Heap<ExtBinTree<char>> hp;
	size_t n=m.size();
	vector<ExtBinTree<char>> fr;
	ExtBinTree<char> Node[hp.DefaultSize];
	if(n>hp.DefaultSize)
	{
		cerr<<"Out of space!"<<endl;
		return NULL;
		exit(1);
	}
	map<char,int>::iterator it=m.begin();
	int i=0;
	while(it!=m.end())
	{
		Node[i].root->c.data=it->first;
		Node[i].root->c.key=it->second;
		Node[i].root->SetLeft(NULL);
		Node[i].root->SetRight(NULL);
		fr.push_back(Node[i]);
		i++;
		it++;
	}
	hp.MakeHeap(fr);
	for(size_t i=0;i<n-1;i++)
	{
		hp.RemoveMin(first);
		hp.RemoveMin(second);
		newtree=new ExtBinTree<char>(first,second);
		hp.Insert(*newtree);
	}
	return newtree;
}

void TextQuery::Encode()
{
	Element<char>* current;
	Element<char>* currentpre;
	size_t length=m.size();
	string substr;
	pair<char,string> p;
	for(size_t i=0;i<length;i++)
	{
		current=currentpre=tree->root;
		substr="";
		while(1)
		{
			if(current->GetLeft()->GetMark()==0)
			{
				substr+='0';
				if(current->GetLeft()->GetLeft()==NULL)
				{
					p=make_pair(current->GetLeft()->c.data,substr);
					current->GetLeft()->SetMark(1);
					this->huffumancode.insert(p);
					break;
				}
				currentpre=current;
				current=current->GetLeft();
			}
			else
			{
				substr+='1';
				if(current->GetRight()->GetLeft()==NULL)
				{
					p=make_pair(current->GetRight()->c.data,substr);
					current->GetRight()->SetMark(1);
					current->SetMark(1);
					huffumancode.insert(p);
					while(currentpre->GetLeft()->GetMark()==1&&currentpre->GetRight()->GetMark()==1)
					{
						currentpre->SetMark(1);
						if(currentpre==tree->root)
							break;
						else if(currentpre->GetParent()!=NULL)
							currentpre=currentpre->GetParent();
						else
							break;
					}
					break;
				}
				currentpre=current;
				current=current->GetRight();
			}
		}
	}
}

int TextQuery::Decode(char& ch,string& str,string file_name)
{
	ofstream on;
	on.open(file_name.c_str());
	if(!on)
	{
		cerr<<"The file cannot open!"<<endl;
		return 0;
	}
	Element<char>* current=tree->root;
	for(size_t i=0;i<str.size();i++)
	{
		if(str[i]=='0')
			current=current->GetLeft();
		else 
			current=current->GetRight();
		if(current->GetLeft()==NULL)
		{
			ch=current->c.data;
			current=tree->root;
			on<<ch;
			cout<<ch;
		}
	}
	cout<<endl;
	on.clear();
	on.close();
	return 1;
}

void TextQuery::Compress(ifstream& in,string source_file,ofstream& on,string destination_file)
{
	on.open("TempCompress.txt",ios::out);
	if(!on)
	{
		cerr<<"The file does not exsit!"<<endl;
		return;
	}
	cout<<strall<<endl;
	for(size_t i=0;i<strall.size();i++)
		str+=huffumancode[strall[i]];
	on<<str;
	strpre=str;
	cout<<str<<endl;
	sum=str.size();
	on.close();
	on.clear();
	in.close();
	in.clear();
	in.open("TempCompress.txt",ios::in);
	if(!in)
	{
		cerr<<"The file does not exsit!"<<endl;
		return;
	}
	on.open(destination_file.c_str(),ios::binary);
	if(!on)
	{
		cerr<<"The file does not exsit!"<<endl;
		return;
	}
	bitset<8> b;
	last=sum%8;
	size_t count=0;
	while(in>>b)
	{
		if(count<sum/8)
		{
			on.write((char*)&b,1);
			cout<<b;
		}
		else
		{
			on<<str.substr(str.size()-last,str.size());
			cout<<str.substr(str.size()-last,str.size());
			break;
		}
		count++;
	}
	cout<<endl;
	in.close();
	in.clear();
	on.close();
	on.clear();
}

void TextQuery::Depress(std::ifstream &in, std::string source_file, std::ofstream &on, std::string destination_file)
{
	in.open(source_file.c_str(),ios::binary);
	if(!in)
	{
		cerr<<"The file does not exsit!"<<endl;
		return;
	}
	on.open("TempDepress.txt",ios::out);
	if(!on)
	{
		cerr<<"The file does not exsit!"<<endl;
		return;
	}
	string substr;
	bitset<8> b;
	size_t count=0;
	int l=sum%8;
	while(in.read((char*)&b,1))
	{
		if(count<sum/8)
		{
			on<<b;
			cout<<b;
		}
		else
		{
			if(l!=0)
			{
				int i=0;
				i=-l;
				in.seekg(i,ios::end);
				in>>substr;
				on<<substr;
				cout<<substr;
			}
			else
				break;
		}
		count++;
	}
	cout<<endl;
	on.clear();
	on.close();
	in.clear();
	in.close();
	in.open("TempDepress.txt",ios::in);
	if(!in)
	{
		cerr<<"The file does not exsit!"<<endl;
		return;
	}
	str="";
	char ch;
	while(in>>ch)
		str+=ch;
	in.clear();
	in.close();
	if(strpre==str)
	{
		Decode(ch,str,destination_file);
	}
	else
		cerr<<"Not the same!"<<endl;
}