lzw.cpp

Implementation for Lempel-Ziv code. Encoder and decoder are separated. The input is a file and th

#pragma warning(disable: 4996)

#include "LZW.h"
#include <iostream>
using namespace std;

LZW::LZW(void)
{
}

LZW::~LZW(void)
{
}

int LZW::Search(const std::list <std::string>& T, const std::string& w)
{
	list<string>::const_iterator it = T.begin();

	int i = 0;
	for(; it != T.end(); it++, i++)
	{
		if (*it == w)
			return i;
	}
	return -1;

}

bool LZW::FindIndex(const std::list <std::string>& T, int i, string& w)
{
	list<string>::const_iterator it = T.begin();

	int j = 0;
	for(; it != T.end(); it++, j++)
	{
		if (j == i)
		{
			w = *it;
			return true;
		}
	}
	return false;
}
void LZW::Compress(const std::string &sInput, const std::string &sOutput)
{
	// open file
	int fdIn = open(sInput.c_str(), O_RDONLY | O_BINARY);
	if (fdIn == -1)
	{
		perror("Error at opening file");
		return;
	}
	int fdOut = open(sOutput.c_str(), O_CREAT | O_TRUNC | O_WRONLY | O_BINARY, _S_IREAD | _S_IWRITE);

	if (fdOut == -1)
	{
		perror("Error at creating/open file");
		close(fdIn);
		return;
	}
	//initialize
	list <string> lT;
	string w = "";
	for(int j = 0; j < SYMBOL_MAX + 1; j++)
	{
		string aux = "";
		aux += (SYMBOL_TYPE)j;
		lT.push_back(aux);
	}

	//compress
	int i = 0;
	bool sourceHasSymbols = true;
	SYMBOL_TYPE c;
	string ww = "";
	while (sourceHasSymbols)
	{
		if (read(fdIn, &c, sizeof(c)) <= 0) 
			sourceHasSymbols = false;
		w = ww;
		ww +=c;
		if ((i = Search(lT, ww)) >= 0)
			continue;
		lT.push_back(ww);
		i = Search(lT, w);
		write(fdOut, &i, BYTES_LIMIT);
		ww = c;
	}
	close(fdOut);
	close(fdIn);
}

void LZW::Decompress(const std::string &sInput, const std::string &sOutput)
{
	// open file
	int fdIn = open(sInput.c_str(), O_RDONLY | O_BINARY);
	if (fdIn == -1)
	{
		perror("Error at opening file");
		return;
	}

	int fdOut = open(sOutput.c_str(), O_CREAT | O_TRUNC | O_WRONLY | O_BINARY, _S_IREAD | _S_IWRITE);
	if (fdOut == -1)
	{
		perror("Error at creating/open file");
		close(fdIn);
		return;
	}

	//initialize
	list <string> lT;
	for(int j = 0; j < SYMBOL_MAX + 1; j++)
	{
		string aux = "";
		aux += (SYMBOL_TYPE)j;
		lT.push_back(aux);
	}

	int cW = 0;
	int pW = 0;
	bool sourceHasSymbols = true;
	string sW;
	string sP;
	string aux;
	SYMBOL_TYPE c;
	read(fdIn, &cW, BYTES_LIMIT);
	FindIndex(lT, cW, sW);
	write(fdOut, sW.c_str(), sW.length());
	while(sourceHasSymbols)
	{
		pW = cW;
		if (read(fdIn, &cW, BYTES_LIMIT) <= 0)
		{
			sourceHasSymbols = false;
			break;
		}

		FindIndex(lT, pW, sP);
		aux = sP;
		if (FindIndex(lT, cW, sW))
		{
			write(fdOut, sW.c_str(), sW.length());
			c = sW[0];
			aux += c;
		}
		else
		{
			c = sP[0];
			aux += c;
			write(fdOut, aux.c_str(), aux.length());
		}
		lT.push_back(aux);
	}

	close(fdIn);
	close(fdOut);
}