6502/lzsa
1
0
Fork 0
mirror of https://github.com/emmanuel-marty/lzsa.git synced 2024-06-01 06:41:43 +00:00
Code Issues Projects Releases Wiki Activity

Increase LZSA2 ratio for some input files

Browse Source
This commit is contained in:
Emmanuel Marty 2020-08-18 09:13:54 +02:00 committed by GitHub
parent 028007b57c
commit 5a0da16874
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 128 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

96
src/shrink_block_v2.c
View File

@ -188,8 +188,8 @@ static inline int lzsa_write_match_varlen_v2(unsigned char *pOutData, int nOutOf
static void lzsa_insert_forward_match_v2(lzsa_compressor *pCompressor, const unsigned char *pInWindow, const int i, const int nMatchOffset, const int nStartOffset, const int nEndOffset, int nDepth) {
lzsa_arrival *arrival = pCompressor->arrival + ((i - nStartOffset) << ARRIVALS_PER_POSITION_SHIFT);
const int *rle_end = (int*)pCompressor->intervals /* reuse */;
lzsa_match* visited = ((lzsa_match*)pCompressor->pos_data) - nStartOffset /* reuse */;
int j;
int nPrevRepPos = -1, nPrevPrevRepPos = -1;
for (j = 0; j < NARRIVALS_PER_POSITION_V2_BIG && arrival[j].from_slot; j++) {
int nRepOffset = arrival[j].rep_offset;
@ -202,15 +202,17 @@ static void lzsa_insert_forward_match_v2(lzsa_compressor *pCompressor, const uns
(nRepPos + nRepLen) <= nEndOffset &&
pCompressor->match[((nRepPos - nStartOffset) << MATCHES_PER_INDEX_SHIFT_V2) + NMATCHES_PER_INDEX_V2 - 1].length == 0) {
if (nPrevRepPos != nRepPos && nPrevPrevRepPos != nRepPos) {
if (visited[nRepPos].offset != nMatchOffset || visited[nRepPos].length > nRepLen) {
visited[nRepPos].offset = nMatchOffset;
visited[nRepPos].length = nRepLen;
if (pInWindow[nRepPos] == pInWindow[nRepPos - nMatchOffset]) {
int nLen0 = rle_end[nRepPos - nMatchOffset] - (nRepPos - nMatchOffset);
int nLen1 = rle_end[nRepPos] - (nRepPos);
int nMinLen = (nLen0 < nLen1) ? nLen0 : nLen1;
if (nMinLen >= nRepLen || !memcmp(pInWindow + nRepPos + nMinLen, pInWindow + nRepPos + nMinLen - nMatchOffset, nRepLen - nMinLen)) {
nPrevPrevRepPos = nPrevRepPos;
nPrevRepPos = nRepPos;
visited[nRepPos].length = 0;
lzsa_match* fwd_match = pCompressor->match + ((nRepPos - nStartOffset) << MATCHES_PER_INDEX_SHIFT_V2);
int r;
@ -267,7 +269,9 @@ static void lzsa_insert_forward_match_v2(lzsa_compressor *pCompressor, const uns
*/
static void lzsa_optimize_forward_v2(lzsa_compressor *pCompressor, const unsigned char *pInWindow, lzsa_match *pBestMatch, const int nStartOffset, const int nEndOffset, const int nReduce, const int nInsertForwardReps, const int nArrivalsPerPosition) {
lzsa_arrival *arrival = pCompressor->arrival - (nStartOffset << ARRIVALS_PER_POSITION_SHIFT);
const int* rle_end = (int*)pCompressor->intervals /* reuse */;
const int *rle_end = (int*)pCompressor->intervals /* reuse */;
lzsa_match *visited = ((lzsa_match*)pCompressor->pos_data) - nStartOffset /* reuse */;
char *nRepLenHandledMask = pCompressor->rep_handled_mask;
const int nModeSwitchPenalty = (pCompressor->flags & LZSA_FLAG_FAVOR_RATIO) ? 0 : MODESWITCH_PENALTY;
const int nMinMatchSize = pCompressor->min_match_size;
const int nDisableScore = nReduce ? 0 : (2 * BLOCK_SIZE);
@ -285,6 +289,10 @@ static void lzsa_optimize_forward_v2(lzsa_compressor *pCompressor, const unsigne
arrival[nStartOffset << ARRIVALS_PER_POSITION_SHIFT].from_slot = -1;
if (nInsertForwardReps) {
memset(visited + nStartOffset, 0, (nEndOffset - nStartOffset) * sizeof(lzsa_match));
}
for (i = nStartOffset; i != nEndOffset; i++) {
lzsa_arrival *cur_arrival = &arrival[i << ARRIVALS_PER_POSITION_SHIFT];
int m;
@ -418,6 +426,10 @@ static void lzsa_optimize_forward_v2(lzsa_compressor *pCompressor, const unsigne
}
}
if (!nReduce) {
memset(nRepLenHandledMask, 0, nArrivalsPerPosition * ((LCP_MAX + 1) / 8) * sizeof(char));
}
for (m = 0; m < NMATCHES_PER_INDEX_V2 && match[m].length; m++) {
int nMatchLen = match[m].length & 0x7fff;
int nMatchOffset = match[m].offset;
@ -546,7 +558,8 @@ static void lzsa_optimize_forward_v2(lzsa_compressor *pCompressor, const unsigne
if (k > nMinOverallRepLen && k <= nMaxOverallRepLen) {
for (j = 0; j < nNumArrivalsForThisPos; j++) {
if (nRepLenForArrival[j] >= k) {
int nMaskOffset = (j << 7) + (k >> 3);
if (nRepLenForArrival[j] >= k && (nReduce || !(nRepLenHandledMask[nMaskOffset] & (1 << (k & 7))))) {
const int nPrevCost = cur_arrival[j].cost & 0x3fffffff;
int nRepCodingChoiceCost = nPrevCost /* the actual cost of the literals themselves accumulates up the chain */ + nMatchLenCost;
int nScore = cur_arrival[j].score + 2;
@ -561,6 +574,8 @@ static void lzsa_optimize_forward_v2(lzsa_compressor *pCompressor, const unsigne
n++) {
if (pDestSlots[n].rep_offset == nRepOffset) {
exists = 1;
if (!nReduce)
nRepLenHandledMask[nMaskOffset] |= 1 << (k & 7);
break;
}
}
@ -1269,8 +1284,77 @@ int lzsa_optimize_and_write_block_v2(lzsa_compressor *pCompressor, const unsigne
nReducedCompressedSize = lzsa_get_compressed_size_v2(pCompressor, pCompressor->improved_match - nPreviousBlockSize, nPreviousBlockSize, nPreviousBlockSize + nInDataSize);
if (nReducedCompressedSize > 0 && nReducedCompressedSize <= nBaseCompressedSize) {
const int nEndOffset = nPreviousBlockSize + nInDataSize;
int nSupplementedCompressedSize;
/* Pick the parse with the reduced number of tokens as it didn't negatively affect the size */
pBestMatch = pCompressor->improved_match - nPreviousBlockSize;
int* first_offset_for_byte = pCompressor->first_offset_for_byte;
int* next_offset_for_pos = pCompressor->next_offset_for_pos;
int nPosition;
/* Supplement small matches */
memset(first_offset_for_byte, 0xff, sizeof(int) * 65536);
memset(next_offset_for_pos, 0xff, sizeof(int) * nInDataSize);
for (nPosition = nPreviousBlockSize; nPosition < nEndOffset - 1; nPosition++) {
next_offset_for_pos[nPosition - nPreviousBlockSize] = first_offset_for_byte[((unsigned int)pInWindow[nPosition]) | (((unsigned int)pInWindow[nPosition + 1]) << 8)];
first_offset_for_byte[((unsigned int)pInWindow[nPosition]) | (((unsigned int)pInWindow[nPosition + 1]) << 8)] = nPosition;
}
for (nPosition = nPreviousBlockSize + 1; nPosition < (nEndOffset - 1); nPosition++) {
lzsa_match* match = pCompressor->match + ((nPosition - nPreviousBlockSize) << MATCHES_PER_INDEX_SHIFT_V2);
int m = 0, nInserted = 0;
int nMatchPos;
while (m < 15 && match[m].length)
m++;
for (nMatchPos = next_offset_for_pos[nPosition - nPreviousBlockSize]; m < 15 && nMatchPos >= 0; nMatchPos = next_offset_for_pos[nMatchPos - nPreviousBlockSize]) {
int nMatchOffset = nPosition - nMatchPos;
int nExistingMatchIdx;
int nAlreadyExists = 0;
for (nExistingMatchIdx = 0; nExistingMatchIdx < m; nExistingMatchIdx++) {
if (match[nExistingMatchIdx].offset == nMatchOffset) {
nAlreadyExists = 1;
break;
}
}
if (!nAlreadyExists) {
int nMatchLen = 2;
while (nMatchLen < 16 && (nPosition + nMatchLen + 4) < nEndOffset && !memcmp(pInWindow + nMatchPos + nMatchLen, pInWindow + nPosition + nMatchLen, 4))
nMatchLen += 4;
while (nMatchLen < 16 && (nPosition + nMatchLen) < nEndOffset && pInWindow[nMatchPos + nMatchLen] == pInWindow[nPosition + nMatchLen])
nMatchLen++;
match[m].length = nMatchLen;
match[m].offset = nMatchOffset;
m++;
nInserted++;
if (nInserted >= 15)
break;
}
}
}
/* Compress optimally with the extra matches */
memset(pCompressor->best_match, 0, BLOCK_SIZE * sizeof(lzsa_match));
lzsa_optimize_forward_v2(pCompressor, pInWindow, pCompressor->best_match - nPreviousBlockSize, nPreviousBlockSize, nPreviousBlockSize + nInDataSize, 1 /* reduce */, 0 /* use forward reps */, nArrivalsPerPosition);
nPasses = 0;
do {
nDidReduce = lzsa_optimize_command_count_v2(pCompressor, pInWindow, pCompressor->best_match - nPreviousBlockSize, nPreviousBlockSize, nPreviousBlockSize + nInDataSize);
nPasses++;
} while (nDidReduce && nPasses < 20);
nSupplementedCompressedSize = lzsa_get_compressed_size_v2(pCompressor, pCompressor->best_match - nPreviousBlockSize, nPreviousBlockSize, nPreviousBlockSize + nInDataSize);
if (nSupplementedCompressedSize > 0 && nSupplementedCompressedSize < nReducedCompressedSize) {
/* Pick the parse with the extra matches as it didn't negatively affect the size */
pBestMatch = pCompressor->best_match - nPreviousBlockSize;
}
}
}

37
src/shrink_context.c
View File

@ -62,6 +62,9 @@ int lzsa_compressor_init(lzsa_compressor *pCompressor, const int nMaxWindowSize,
pCompressor->best_match = NULL;
pCompressor->improved_match = NULL;
pCompressor->arrival = NULL;
pCompressor->rep_handled_mask = NULL;
pCompressor->first_offset_for_byte = NULL;
pCompressor->next_offset_for_pos = NULL;
pCompressor->min_match_size = nMinMatchSize;
if (pCompressor->min_match_size < nMinMatchSizeForFormat)
pCompressor->min_match_size = nMinMatchSizeForFormat;
@ -102,8 +105,23 @@ int lzsa_compressor_init(lzsa_compressor *pCompressor, const int nMaxWindowSize,
pCompressor->match = (lzsa_match *)malloc(BLOCK_SIZE * NMATCHES_PER_INDEX_V2 * sizeof(lzsa_match));
else
pCompressor->match = (lzsa_match *)malloc(BLOCK_SIZE * NMATCHES_PER_INDEX_V1 * sizeof(lzsa_match));
if (pCompressor->match)
return 0;
if (pCompressor->match) {
if (pCompressor->format_version == 2) {
pCompressor->rep_handled_mask = (char*)malloc(NARRIVALS_PER_POSITION_V2_BIG * ((LCP_MAX + 1) / 8) * sizeof(char));
if (pCompressor->rep_handled_mask) {
pCompressor->first_offset_for_byte = (int*)malloc(65536 * sizeof(int));
if (pCompressor->first_offset_for_byte) {
pCompressor->next_offset_for_pos = (int*)malloc(BLOCK_SIZE * sizeof(int));
if (pCompressor->next_offset_for_pos) {
return 0;
}
}
}
}
else {
return 0;
}
}
}
}
}
@ -124,6 +142,21 @@ int lzsa_compressor_init(lzsa_compressor *pCompressor, const int nMaxWindowSize,
void lzsa_compressor_destroy(lzsa_compressor *pCompressor) {
divsufsort_destroy(&pCompressor->divsufsort_context);
if (pCompressor->next_offset_for_pos) {
free(pCompressor->next_offset_for_pos);
pCompressor->next_offset_for_pos = NULL;
}
if (pCompressor->first_offset_for_byte) {
free(pCompressor->first_offset_for_byte);
pCompressor->first_offset_for_byte = NULL;
}
if (pCompressor->rep_handled_mask) {
free(pCompressor->rep_handled_mask);
pCompressor->rep_handled_mask = NULL;
}
if (pCompressor->match) {
free(pCompressor->match);
pCompressor->match = NULL;

3
src/shrink_context.h
View File

@ -124,6 +124,9 @@ typedef struct _lzsa_compressor {
lzsa_match *best_match;
lzsa_match *improved_match;
lzsa_arrival *arrival;
char *rep_handled_mask;
int *first_offset_for_byte;
int *next_offset_for_pos;
int min_match_size;
int format_version;
int flags;