190 lines
5.3 KiB
C
190 lines
5.3 KiB
C
#include <assert.h>
|
|
#include <stdio.h>
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
#include <ctype.h>
|
|
#include "compressor.h"
|
|
#define BUFSIZE 1024
|
|
|
|
|
|
char* leftRotatebyOne(char arr[BUFSIZE * 2]){
|
|
char* result = (char*) calloc(strlen(arr), sizeof(char));
|
|
char temp = arr[0];
|
|
for (int i = 0; i < strlen(arr) - 1; i++)
|
|
arr[i] = arr[i + 1];
|
|
arr[strlen(arr)-1] = temp;
|
|
strcpy(result, arr);
|
|
return result;
|
|
}
|
|
|
|
|
|
char* bwt_compress(char buffer[BUFSIZE * 2]){
|
|
char* encoded = (char*) calloc(BUFSIZE * 2, sizeof(char));
|
|
char rotate_temp[BUFSIZE * 2];
|
|
for(int i = 0; i < strlen(buffer); i++)
|
|
rotate_temp[i] = buffer[i];
|
|
rotate_temp[strlen(buffer)] = '\0';
|
|
char* rotation[strlen(buffer)];
|
|
for(int i = 0; i < strlen(buffer); i++)
|
|
rotation[i] = (char*) calloc(BUFSIZE * 2, sizeof(char));
|
|
|
|
for(int i = 0; i < strlen(buffer); i++)
|
|
strcpy(rotation[i], leftRotatebyOne(rotate_temp));
|
|
|
|
char* temp = (char*) calloc(BUFSIZE * 2, sizeof(char));
|
|
for(int i = 0; i < strlen(buffer); i++){
|
|
for(int j = 0; j < strlen(buffer) - i - 1; j++){
|
|
if(strcmp(rotation[j], rotation[j+1]) > 0){
|
|
strcpy(temp, rotation[j]);
|
|
strcpy(rotation[j], rotation[j+1]);
|
|
strcpy(rotation[j+1], temp);
|
|
}
|
|
}
|
|
}
|
|
|
|
|
|
for(int i = 0; i < strlen(buffer); i++) {
|
|
strncat(encoded, &rotation[i][strlen(rotation[i]) - 1], 1);
|
|
if(strcmp(rotation[i], buffer) == 0) {
|
|
char str[100];
|
|
for(int j = 0; j < 100; j++)
|
|
str[j] = '\0';
|
|
sprintf(str, "%d", i);
|
|
FILE* num = fopen("num.txt", "w");
|
|
fwrite(str, sizeof(char), strlen(str), num);
|
|
}
|
|
}
|
|
|
|
return encoded;
|
|
}
|
|
|
|
|
|
char* run_length(char buffer[BUFSIZE]){
|
|
char* encoded = (char*) calloc(BUFSIZE * 2, sizeof(char));
|
|
int counter = 1;
|
|
int i = 1;
|
|
|
|
while(i < strlen(buffer)+1){
|
|
for(; i < strlen(buffer)+1; i++){
|
|
if(buffer[i] == buffer[i-1])
|
|
counter++;
|
|
else {
|
|
char* temp = (char*) calloc(5, sizeof(char));
|
|
sprintf(temp, "%d", counter);
|
|
strncat(encoded, temp, strlen(temp));
|
|
strncat(encoded, &buffer[i-1], 1);
|
|
i++;
|
|
counter = 1;
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
return encoded;
|
|
}
|
|
|
|
|
|
void compress(FILE* infile,FILE* outfile){
|
|
char buffer[BUFSIZE];
|
|
memset(buffer,0,BUFSIZE);
|
|
while(1) {
|
|
int insize = (int) fread(buffer, sizeof(char), BUFSIZE, infile);
|
|
if (insize == 0) {
|
|
if (feof(infile)) {
|
|
// end of file
|
|
break;
|
|
}
|
|
assert(!ferror(infile));
|
|
}
|
|
|
|
char outbuf[BUFSIZE*2];
|
|
strncpy(outbuf, run_length(bwt_compress(buffer)), BUFSIZE*2-1);
|
|
|
|
fwrite(outbuf,sizeof(char), strlen(outbuf),outfile);
|
|
}
|
|
}
|
|
|
|
|
|
char* bwt_decompress(char buffer[BUFSIZE * 2]){
|
|
char* decoded = (char*) calloc(BUFSIZE * 2, sizeof(char));
|
|
char rotation[strlen(buffer)][BUFSIZE * 2];
|
|
for(int i = 0; i < strlen(buffer); i++) {
|
|
for (int j = 0; j < strlen(buffer) + 1; j++) {
|
|
if(j != strlen(buffer))
|
|
rotation[i][j] = ' ';
|
|
else
|
|
rotation[i][j] = '\0';
|
|
}
|
|
}
|
|
|
|
for(int i = 0; i < strlen(buffer); i++) {
|
|
for (int j = 0; j < strlen(buffer) + 1; j++) {
|
|
rotation[j][strlen(buffer) - 1 - i] = buffer[j];
|
|
}
|
|
|
|
char *temp = (char *) calloc(BUFSIZE * 2, sizeof(char));
|
|
for (int iterator = 0; iterator < strlen(buffer); iterator++) {
|
|
for (int j = 0; j < strlen(buffer) - 1 - iterator; j++) {
|
|
if (strcmp(rotation[j], rotation[j + 1]) > 0) {
|
|
strcpy(temp, rotation[j]);
|
|
strcpy(rotation[j], rotation[j + 1]);
|
|
strcpy(rotation[j + 1], temp);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
FILE* num = fopen("num.txt", "r");
|
|
char number[100];
|
|
for(int i = 0; i < 100; i++)
|
|
number[i] = '\0';
|
|
fread(number, sizeof(char), 100, num);
|
|
strcpy(decoded, rotation[atoi(number)]);
|
|
|
|
return decoded;
|
|
}
|
|
|
|
|
|
char* run_length_decompress(char buffer[BUFSIZE * 2]){
|
|
char* decoded = (char*) calloc(BUFSIZE, sizeof(char));
|
|
int i = 0;
|
|
|
|
while(i < strlen(buffer)){
|
|
char* temp = (char*) calloc(5, sizeof(char));
|
|
for(; i < strlen(buffer); i++){
|
|
if(isdigit(buffer[i])){
|
|
strncat(temp, &buffer[i], 1);
|
|
}
|
|
else{
|
|
for(int j = 0; j < atoi(temp); j++)
|
|
strncat(decoded, &buffer[i], 1);
|
|
i++;
|
|
break;
|
|
}
|
|
}
|
|
memset(temp, '\0', 5);
|
|
}
|
|
|
|
return decoded;
|
|
}
|
|
|
|
|
|
void decompress(FILE* infile,FILE* outfile){
|
|
char buffer[BUFSIZE * 2];
|
|
memset(buffer, 0, BUFSIZE*2);
|
|
while(1){
|
|
int insize = (int)fread(buffer,sizeof(char),BUFSIZE * 2,infile);
|
|
if (insize == 0){
|
|
if (feof(infile)){
|
|
// end of file
|
|
break;
|
|
}
|
|
assert(!ferror(infile));
|
|
}
|
|
char outbuf[BUFSIZE];
|
|
strncpy(outbuf, bwt_decompress(run_length_decompress(buffer)), BUFSIZE-1);
|
|
|
|
fwrite(outbuf,sizeof(char), strlen(outbuf),outfile);
|
|
}
|
|
}
|