int lz77_compress(const unsigned char *original, unsigned char **compressed,
int size)
{
unsigned long token;
unsigned char window[LZ77_WINDOW_SIZE],
buffer[LZ77_BUFFER_SIZE], *comp, *temp, next;
int offset, length, remaining, tbits, hsize, ipos, opos, tpos, i;
/*****************************************************************************
* Make the pointer to the compressed data not valid until later. *
*****************************************************************************/
*compressed = NULL;
/*****************************************************************************
* Write the header information. *
*****************************************************************************/
hsize = sizeof(int);
if ((comp = (unsigned char *)malloc(hsize)) == NULL)
return -1;
memcpy(comp, &size, sizeof(int));
/*****************************************************************************
* Initialize the sliding window and the look-ahead buffer. *
*****************************************************************************/
memset(window, 0, LZ77_WINDOW_SIZE);
memset(buffer, 0, LZ77_BUFFER_SIZE);
/*****************************************************************************
* Load the look-ahead buffer. *
*****************************************************************************/
ipos = 0;
for (i = 0; i < LZ77_BUFFER_SIZE && ipos < size; i++) {
buffer[i] = original[ipos];
ipos++;
}
/*****************************************************************************
* Compress the data. *
*****************************************************************************/
opos = hsize * 8;
remaining = size;
while (remaining > 0) {
if ((length = compare_win(window, buffer, &offset, &next)) != 0) {
/***********************************************************************
* Encode a phrase token. *
***********************************************************************/
token = 0x00000001 << (LZ77_PHRASE_BITS - 1);
/***********************************************************************
* Set the offset where the match was found in the sliding window. *
***********************************************************************/
token =
token | (offset <<
(LZ77_PHRASE_BITS - LZ77_TYPE_BITS -
LZ77_WINOFF_BITS));
/***********************************************************************
* Set the length of the match. *
***********************************************************************/
token =
token | (length <<
(LZ77_PHRASE_BITS - LZ77_TYPE_BITS -
LZ77_WINOFF_BITS - LZ77_BUFLEN_BITS));
/***********************************************************************
* Set the next symbol in the look-ahead buffer after the match. *
***********************************************************************/
token = token | next;
/***********************************************************************
* Set the number of bits in the token. *
***********************************************************************/
tbits = LZ77_PHRASE_BITS;
}
else {
/***********************************************************************
* Encode a symbol token. *
***********************************************************************/
token = 0x00000000;
/***********************************************************************
* Set the unmatched symbol. *
***********************************************************************/
token = token | next;
/***********************************************************************
* Set the number of bits in the token. *
***********************************************************************/
tbits = LZ77_SYMBOL_BITS;
}
/**************************************************************************
* Ensure that the token is in big-endian format. *
**************************************************************************/
token = htonl(token);
/**************************************************************************
* Write the token to the buffer of compressed data. *
**************************************************************************/
for (i = 0; i < tbits; i++) {
if (opos % 8 == 0) {
/********************************************************************
* Allocate another byte for the buffer of compressed data. *
********************************************************************/
if ((temp =
(unsigned char *)realloc(comp,
(opos / 8) +
1)) == NULL) {
free(comp);
return -1;
}
comp = temp;
}
tpos = (sizeof(unsigned long) * 8) - tbits + i;
bit_set(comp, opos,
bit_get((unsigned char *)&token, tpos));
opos++;
}
/**************************************************************************
* Adjust the phrase length to account for the unmatched symbol. *
**************************************************************************/
length++;
/**************************************************************************
* Copy data from the look-ahead buffer to the sliding window. *
**************************************************************************/
memmove(&window[0], &window[length], LZ77_WINDOW_SIZE - length);
memmove(&window[LZ77_WINDOW_SIZE - length], &buffer[0], length);
/**************************************************************************
* Read more data into the look-ahead buffer. *
**************************************************************************/
memmove(&buffer[0], &buffer[length], LZ77_BUFFER_SIZE - length);
for (i = LZ77_BUFFER_SIZE - length;
i < LZ77_BUFFER_SIZE && ipos < size; i++) {
buffer[i] = original[ipos];
ipos++;
}
/**************************************************************************
* Adjust the total symbols remaining by the phrase length. *
**************************************************************************/
remaining = remaining - length;
}
/*****************************************************************************
* Point to the buffer of compressed data. *
*****************************************************************************/
*compressed = comp;
/*****************************************************************************
* Return the number of bytes in the compressed data. *
*****************************************************************************/
return ((opos - 1) / 8) + 1;
}
int lz77_compress(const unsigned char *original, unsigned char **compressed, int size) {
unsigned char window[LZ77_WINDOW_SIZE], buffer[LZ77_BUFFER_SIZE], *comp, *temp, next;
int offset, length, remaining, hsize, ipos, opos, tpos, i, tbits, token;
/* Make pointer to compressed data not valid until later */
*compressed = NULL;
/* Write header info */
hsize = sizeof(int);
if ((comp = (unsigned char *) malloc(hsize)) == NULL)
return -1;
memcpy(comp, &size, sizeof(int));
/* Initialize the sliding window and look-ahead buffer */
memset(window, 0, LZ77_WINDOW_SIZE);
memset(buffer, 0, LZ77_BUFFER_SIZE);
/* Load look-ahead buffer */
ipos = 0;
for (i = 0; i < LZ77_BUFFER_SIZE && ipos < size; i++) {
buffer[i] = original[ipos];
ipos++;
}
/* Compress data */
opos = hsize * 8;
remaining = size;
while (remaining > 0) {
if ((length = compare_win(window, buffer, &offset, &next)) != 0) {
if (DEBUG) printf("PHRASE TOKEN\n");
/* Encode a phrase token */
token = 0x00000001;
dump("token before", token, sizeof(int) * 8);
token = 0x00000001 << (LZ77_PHRASE_BITS - LZ77_TYPE_BITS);
dump("token", token, sizeof(int) * 8);
/* Set offset where match was found in the sliding window */
dump("offset", offset, sizeof(int) * 8);
token = token | (offset << (LZ77_PHRASE_BITS - LZ77_TYPE_BITS - LZ77_WINOFF_BITS));
int offset_shifted = offset << (LZ77_PHRASE_BITS - LZ77_TYPE_BITS - LZ77_WINOFF_BITS);
if (DEBUG) printf("bitshift = %d\n", LZ77_PHRASE_BITS - LZ77_TYPE_BITS - LZ77_WINOFF_BITS);
dump("offset after bitshift", offset_shifted, sizeof(int) * 8);
dump("token", token, sizeof(int) * 8);
/* Set length of match */
token = token | (length << (LZ77_PHRASE_BITS - LZ77_TYPE_BITS - LZ77_WINOFF_BITS - LZ77_BUFLEN_BITS));
dump("length", length, LZ77_WINOFF_BITS);
dump("token", token, sizeof(int) * 8);
/* Set next symbol in the look-ahead buffer after the match */
token = token | next;
dump("next", next, sizeof(char) * 8);
/* Set number of bits in the token */
tbits = LZ77_PHRASE_BITS;
dump("token", token, sizeof(int) * 8);
} else {
if (DEBUG) printf("SYMBOL TOKEN\n");
/* Encode a symbol token */
token = 0x00000000;
/* Set the unmatched symbol */
token = token | next;
dump("next", next, sizeof(char) * 8);
/* Set number of bits in the token */
tbits = LZ77_SYMBOL_BITS;
dump("token", token, sizeof(int) * 8);
}
/* Ensure that the token is in big-endian format */
token = htonl(token);
dump("token big endian", token, sizeof(int) * 8);
/* Write the token to the buffer of compressed data */
if (DEBUG) printf("tbits = %d\n", tbits);
for (i = 0; i < tbits; i++) {
if (opos % 8 == 0) {
/* Allocate another byte for buffer of compressed data */
if ((temp = (unsigned char *) realloc(comp, (opos / 8) + 1)) == NULL) {
free(comp);
return -1;
}
comp = temp;
}
/* Book had sizeof(unsigned long) which I think is wrong */
/* Calculating appropriate tpos for big endian format */
tpos = (sizeof(int) * 8) - tbits + i;
if (DEBUG) printf("tpos = %d, opos = %d\n", tpos, opos);
bit_set(comp, opos, bit_get((unsigned char *) &token, tpos));
if (DEBUG) {
printf("comp = ");
int j;
for (j = 0; j <= opos; j++) {
if (j % 4 == 0) printf(" ");
printf("%d", bit_get(comp, j));
}
printf("\n");
}
opos++;
}
/* Adjust the phrase length to account for unmatched symbol */
length++;
/* Copy data from look-ahead buffer to sliding window */
memmove(&window[0], &window[length], LZ77_WINDOW_SIZE - length);
memmove(&window[LZ77_WINDOW_SIZE - length], &buffer[0], length);
/* Read more data into look-ahead buffer */
memmove(&buffer[0], &buffer[length], LZ77_BUFFER_SIZE - length);
for (i = LZ77_BUFFER_SIZE - length; i < LZ77_BUFFER_SIZE && ipos < size; i++) {
buffer[i] = original[ipos];
ipos++;
}
/* Adjust total symbols remaining by phrase length */
remaining = remaining - length;
}
/* Point to the buffer of compressed data */
*compressed = comp;
/* Return the number of bytes in compressed data */
return ((opos - 1) / 8) + 1;
}