8.8 KiB
#include <assert.h> #include <stdio.h> #include <stdlib.h> #include <string.h> #include "compressor.h" #define BUFSIZE 1024 #define SIZE 256 #define MAX_WORDS 100000
// One codeword is previous index - new index -character
// Trie (a.k.a prefix tree) struct dict { int index; // Character is encoded as non zero value in characters struct dict* characters[SIZE]; };
// Free trie void delete(struct dict* tree){ //skopirovane z lz78, vymaze strukturu co sa pouziva v lz78 if (tree == NULL){ //maze to pamat stromu return; } for (int i = 0 ;i < SIZE; i++){ delete(tree-> characters[i]); } }
void get_words (struct dict* tree,char* words[]){ //posielam cely strom a smernik na dvojrozmerne pole if (tree == NULL){ return; } for (int i = 0 ;i < SIZE; i++){ //prejdeme vsetkych potomkov if (tree-> characters[i] > 0){ //zistime ci tam je nieco naalokovane int need_size = snprintf(NULL,0,"-%d-%c\n",tree->index,i); //kolko pamate words[tree->characters[i]->index] = calloc(need_size + 1,sizeof(char)); //naalokujeme pamat memset(words[tree->characters[i]->index],0,need_size + 1); sprintf(words[tree->characters[i]->index],"-%d-%c\n",tree->index,i); //napise sa vo formate -index predosleho notu - aktualny znak
get_words(tree-> characters[i],words); //zavolam na kazdeho potomka a potom na dalsich potomkov alebo syna
}
}
}
int mylz78compress(char* buff,int size,char** outbuf){
// https://medium.com/@dbudhrani/how-data-compression-works-exploring-lz78-e97e539138
// The trie represents the current phrase
// The first node represents the empty string with index 0
struct dict* root = calloc(1,sizeof(struct dict));
struct dict* current = root;
int index = 1;
int c = 0;
char *last = NULL;
for(int i = 0; i < size; i++){
c = buff[i];
// Current character is in dictionary, continue searching
if (current->characters[c] > 0){
// ak este nieco ostalo a toto je koniec
if (i == size - 1){ // pridem na koniec suboru a musim si to zapamatat, inac to stratim
int sz_needed = snprintf(NULL,0,"-%d-%c\n",current->index,c); //kolko bajtov to bude zaberat
last = calloc(sz_needed + 1,sizeof(char)); //alokuje do do last
sprintf(last,"-%d-%c\n",current->index,c); //
}
current = current->characters[c];
}
else {
current->characters[c] = calloc(1,sizeof(struct dict));
current->characters[c]->index = index;
index += 1;
current = root;
}
}
char **words = calloc(index,sizeof(char*)); // z stromu vytvori format
get_words(root,words);
int size_needed = 0;
for(int i=1; i < index; i++){ //nastavime velkost aku bude mat vystupny buffer
if(words[i] != NULL){
size_needed += strlen(words[i]); //poscitavame velkosti vsetkych stringov
}
}
if (last != NULL){ //mame string last?
size_needed += strlen(last); //priratam
}
char *temp = calloc(size_needed + 10,sizeof(char)); //alokujem velkost, padalo mi to ked som mal mensie cislo ako 10
int pos = 0; //pamatam kde som to uz vypisal
for(int i=0; i < index; i++){
if (words[i] != NULL){
strcpy(temp + pos,words[i]); //nebude sa to pisat na zaciatok ale az za tym
pos += strlen(words[i]);
}
}
if (last != NULL){ // zapisem aj last
strcpy(temp + pos,last);
pos += strlen(last);
}
delete(root); //nepotrebujeme strom, uvolnime pamat
*outbuf = temp; //adresa sa zapise do outbuffer
return strlen(temp);
}
int myrlcompress(char* buff,int size,char** outbuf){ // 2 * size je urcite dost, kedze v najhorsom pripade bude kazdy znak osamote *outbuf = calloc(2 * size + 1,sizeof(char)); //smernik na smernik, napisem adresu novej pamate memset(*outbuf,0,size + 1); //vymaze
int head = -1;
int length = 0;
int pos = 0;
for(int i = 0; i < size; i++){
int c = buff[i];
if (c == head){
length += 1;
if (i == size - 1){ //posledne pismenko
(*outbuf)[pos] = head;
(*outbuf)[pos + 1] = length;
pos += 2; //aby som vedel kde to zapisovat
break;
}
}
else {
if (head != -1){
(*outbuf)[pos] = head;
(*outbuf)[pos + 1] = length;
pos += 2;
}
if (i == size - 1){ //posledne pismenko
(*outbuf)[pos] = c;
(*outbuf)[pos + 1] = 1;
pos += 2;
break;
}
head = c;
length = 1;
}
// ak sme blizko preteceniu, vypiseme aktualny znak
if (length == 255){ //nemozme ist viac, vynulujem a idem od nuly
(*outbuf)[pos] = head;
(*outbuf)[pos + 1] = length;
pos += 2;
head = -1;
length = 0;
}
}
return pos; //velkost vystupneho buffera
}
void compress(FILE* infile,FILE* outfile){ fseek(infile, 0, SEEK_END); //vstupny subor, a chceme presunut kurzor na konci suboru int insize = ftell(infile) + 1; // na akej pozicii je kurzor, zistime velkost suboru rewind(infile); //vrati kurzor na zaciatok char *buffer = calloc(insize,sizeof(char)); //alokuje pamat
memset(buffer,0,insize);
insize = fread(buffer,sizeof(char),insize - 1,infile); //nacita obsah do bufferu
if (insize == 0){
assert(!ferror(infile));
}
char *tempbuf = NULL; // az vo funkcii to alokujem
int tempsize = myrlcompress(buffer,insize,&tempbuf); //vstupny buffer, velkost a adresu bufferu
char *outbuf = NULL;
int outsize = mylz78compress(tempbuf,tempsize,&outbuf);
if (outsize > 0){
fwrite(outbuf,sizeof(char),outsize,outfile); //napise sa do suboru, z akeho buffera do akeho suboru
}
}
int myrldecompress(char* buff,int size,char** outbuf){ *outbuf = calloc(size + 1,sizeof(char)); memset(*outbuf,0,size + 1);
int pos = 0;
for(int i = 0; i < size; i+=2){
char c = buff[i];
char length = buff[i + 1];
// ak nam nebude stacit miesto v *outbuf, treba allocovat viac
while((length + pos) > (size - 1)){
size *= 2;
char *tempbuf = calloc(size,sizeof(char));
memset(tempbuf,'\0',size);
strcpy(tempbuf,*outbuf);
*outbuf = tempbuf;
}
for (int j = 0; j < length; j++){ //tolko krat ho pridavam, kolko mam cisielko
(*outbuf)[pos] = c;
pos += 1;
}
}
return pos;
}
int mylz78decompress(char* buff,int size,char** outbuf){ //nepotrebujem pouzivat strom *outbuf = calloc(size,sizeof(char)); //taka ista velkost ako vstup memset(*outbuf,0,size); char *words[MAX_WORDS]; //dvojrozmerne pole words[0] = ""; char c; int index = 1; int number; int pos = 0; int bytes_read, bytes_read_all = 0; //citame vstupny buffer, necitame po bajtoch, viac ciferne cisla
while (sscanf(buff + bytes_read_all,"-%d-%c\n%n",&number,&c,&bytes_read) == 2){ //formatovacie citanie, pusaveme o bajty, ktore sa precitali ak vrati ine cislo ako 2 tak je nakocni
bytes_read_all += bytes_read; //o tolko sme sa posunuli
words[index] = calloc(strlen(words[number]) + 2,sizeof(char*)); //vo words sa vytvoria prefixi
strcpy(words[index],words[number]); //nakopirujeme words number a pridame nakoniec c
words[index][strlen(words[number])] = c;
// ak nam nebude stacit miesto v *outbuf, treba allocovat viac
while((strlen(words[index]) + pos) > (size - 1)){
size *= 2;
char *tempbuf = calloc(size,sizeof(char));
memset(tempbuf,'\0',size);
strcpy(tempbuf,*outbuf);
*outbuf = tempbuf;
}
strcpy(*outbuf + pos,words[index]); //nakopiruje najnovsie slovo
pos += strlen(words[index]);
index += 1;
}
return strlen(*outbuf); //vratim velkost
}
void decompress(FILE* infile,FILE* outfile){ fseek(infile, 0, SEEK_END); int insize = ftell(infile) + 1; rewind(infile); char *buffer = calloc(insize,sizeof(char));
memset(buffer,0,insize);
insize = fread(buffer,sizeof(char),insize - 1,infile);
if (insize == 0){
assert(!ferror(infile));
}
char *tempbuf = NULL;
int tempsize = mylz78decompress(buffer,insize,&tempbuf); //najprv lz78
char *outbuf = NULL;
int outsize = myrldecompress(tempbuf,tempsize,&outbuf);
if (outsize > 0){
fwrite(outbuf,sizeof(char),outsize,outfile);
}
}