1
/* deflate.c -- compress data using the deflation algorithm
2
* Copyright (C) 1995-1998 Jean-loup Gailly.
3
* For conditions of distribution and use, see copyright notice in zlib.h
9
* The "deflation" process depends on being able to identify portions
10
* of the input text which are identical to earlier input (within a
11
* sliding window trailing behind the input currently being processed).
13
* The most straightforward technique turns out to be the fastest for
14
* most input files: try all possible matches and select the longest.
15
* The key feature of this algorithm is that insertions into the string
16
* dictionary are very simple and thus fast, and deletions are avoided
17
* completely. Insertions are performed at each input character, whereas
18
* string matches are performed only when the previous match ends. So it
19
* is preferable to spend more time in matches to allow very fast string
20
* insertions and avoid deletions. The matching algorithm for small
21
* strings is inspired from that of Rabin & Karp. A brute force approach
22
* is used to find longer strings when a small match has been found.
23
* A similar algorithm is used in comic (by Jan-Mark Wams) and freeze
24
* (by Leonid Broukhis).
25
* A previous version of this file used a more sophisticated algorithm
26
* (by Fiala and Greene) which is guaranteed to run in linear amortized
27
* time, but has a larger average cost, uses more memory and is patented.
28
* However the F&G algorithm may be faster for some highly redundant
29
* files if the parameter max_chain_length (described below) is too large.
33
* The idea of lazy evaluation of matches is due to Jan-Mark Wams, and
34
* I found it in 'freeze' written by Leonid Broukhis.
35
* Thanks to many people for bug reports and testing.
39
* Deutsch, L.P.,"DEFLATE Compressed Data Format Specification".
40
* Available in ftp://ds.internic.net/rfc/rfc1951.txt
42
* A description of the Rabin and Karp algorithm is given in the book
43
* "Algorithms" by R. Sedgewick, Addison-Wesley, p252.
45
* Fiala,E.R., and Greene,D.H.
46
* Data Compression with Finite Windows, Comm.ACM, 32,4 (1989) 490-595
50
/* @(#) $Id: deflate.c,v 1.1.1.1 2000/11/17 14:22:46 rocco Exp $ */
54
const char deflate_copyright[] =
55
" deflate 1.1.3 Copyright 1995-1998 Jean-loup Gailly ";
57
If you use the zlib library in a product, an acknowledgment is welcome
58
in the documentation of your product. If for some reason you cannot
59
include such an acknowledgment, I would appreciate that you keep this
60
copyright string in the executable of your product.
63
/* ===========================================================================
64
* Function prototypes.
67
need_more, /* block not completed, need more input or more output */
68
block_done, /* block flush performed */
69
finish_started, /* finish started, need only more output at next deflate */
70
finish_done /* finish done, accept no more input or output */
73
typedef block_state (*compress_func) OF((deflate_state *s, int flush));
74
/* Compression function. Returns the block state after the call. */
76
local void fill_window OF((deflate_state *s));
77
local block_state deflate_stored OF((deflate_state *s, int flush));
78
local block_state deflate_fast OF((deflate_state *s, int flush));
79
local block_state deflate_slow OF((deflate_state *s, int flush));
80
local void lm_init OF((deflate_state *s));
81
local void putShortMSB OF((deflate_state *s, uInt b));
82
local void flush_pending OF((z_streamp strm));
83
local int read_buf OF((z_streamp strm, Bytef *buf, unsigned size));
85
void match_init OF((void)); /* asm code initialization */
86
uInt longest_match OF((deflate_state *s, IPos cur_match));
88
local uInt longest_match OF((deflate_state *s, IPos cur_match));
92
local void check_match OF((deflate_state *s, IPos start, IPos match,
96
/* ===========================================================================
101
/* Tail of hash chains */
104
# define TOO_FAR 4096
106
/* Matches of length 3 are discarded if their distance exceeds TOO_FAR */
108
#define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1)
109
/* Minimum amount of lookahead, except at the end of the input file.
110
* See deflate.c for comments about the MIN_MATCH+1.
113
/* Values for max_lazy_match, good_match and max_chain_length, depending on
114
* the desired pack level (0..9). The values given below have been tuned to
115
* exclude worst case performance for pathological files. Better values may be
116
* found for specific files.
118
typedef struct config_s {
119
ush good_length; /* reduce lazy search above this match length */
120
ush max_lazy; /* do not perform lazy search above this match length */
121
ush nice_length; /* quit search above this match length */
126
local const config configuration_table[10] = {
127
/* good lazy nice chain */
128
/* 0 */ {0, 0, 0, 0, deflate_stored}, /* store only */
129
/* 1 */ {4, 4, 8, 4, deflate_fast}, /* maximum speed, no lazy matches */
130
/* 2 */ {4, 5, 16, 8, deflate_fast},
131
/* 3 */ {4, 6, 32, 32, deflate_fast},
133
/* 4 */ {4, 4, 16, 16, deflate_slow}, /* lazy matches */
134
/* 5 */ {8, 16, 32, 32, deflate_slow},
135
/* 6 */ {8, 16, 128, 128, deflate_slow},
136
/* 7 */ {8, 32, 128, 256, deflate_slow},
137
/* 8 */ {32, 128, 258, 1024, deflate_slow},
138
/* 9 */ {32, 258, 258, 4096, deflate_slow}}; /* maximum compression */
140
/* Note: the deflate() code requires max_lazy >= MIN_MATCH and max_chain >= 4
141
* For deflate_fast() (levels <= 3) good is ignored and lazy has a different
146
/* result of memcmp for equal strings */
148
struct static_tree_desc_s {int dummy;}; /* for buggy compilers */
150
/* ===========================================================================
151
* Update a hash value with the given input byte
152
* IN assertion: all calls to to UPDATE_HASH are made with consecutive
153
* input characters, so that a running hash key can be computed from the
154
* previous key instead of complete recalculation each time.
156
#define UPDATE_HASH(s,h,c) (h = (((h)<<s->hash_shift) ^ (c)) & s->hash_mask)
159
/* ===========================================================================
160
* Insert string str in the dictionary and set match_head to the previous head
161
* of the hash chain (the most recent string with same hash key). Return
162
* the previous length of the hash chain.
163
* If this file is compiled with -DFASTEST, the compression level is forced
164
* to 1, and no hash chains are maintained.
165
* IN assertion: all calls to to INSERT_STRING are made with consecutive
166
* input characters and the first MIN_MATCH bytes of str are valid
167
* (except for the last MIN_MATCH-1 bytes of the input file).
170
#define INSERT_STRING(s, str, match_head) \
171
(UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \
172
match_head = s->head[s->ins_h], \
173
s->head[s->ins_h] = (Pos)(str))
175
#define INSERT_STRING(s, str, match_head) \
176
(UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \
177
s->prev[(str) & s->w_mask] = match_head = s->head[s->ins_h], \
178
s->head[s->ins_h] = (Pos)(str))
181
/* ===========================================================================
182
* Initialize the hash table (avoiding 64K overflow for 16 bit systems).
183
* prev[] will be initialized on the fly.
185
#define CLEAR_HASH(s) \
186
s->head[s->hash_size-1] = NIL; \
187
zmemzero((Bytef *)s->head, (unsigned)(s->hash_size-1)*sizeof(*s->head));
189
/* ========================================================================= */
190
int ZEXPORT deflateInit_(strm, level, version, stream_size)
196
return deflateInit2_(strm, level, Z_DEFLATED, MAX_WBITS, DEF_MEM_LEVEL,
197
Z_DEFAULT_STRATEGY, version, stream_size);
198
/* To do: ignore strm->next_in if we use it as window */
201
/* ========================================================================= */
202
int ZEXPORT deflateInit2_(strm, level, method, windowBits, memLevel, strategy,
203
version, stream_size)
215
static const char* my_version = ZLIB_VERSION;
218
/* We overlay pending_buf and d_buf+l_buf. This works since the average
219
* output size for (length,distance) codes is <= 24 bits.
222
if (version == Z_NULL || version[0] != my_version[0] ||
223
stream_size != sizeof(z_stream)) {
224
return Z_VERSION_ERROR;
226
if (strm == Z_NULL) return Z_STREAM_ERROR;
229
if (strm->zalloc == Z_NULL) {
230
strm->zalloc = zcalloc;
231
strm->opaque = (voidpf)0;
233
if (strm->zfree == Z_NULL) strm->zfree = zcfree;
235
if (level == Z_DEFAULT_COMPRESSION) level = 6;
240
if (windowBits < 0) { /* undocumented feature: suppress zlib header */
242
windowBits = -windowBits;
244
if (memLevel < 1 || memLevel > MAX_MEM_LEVEL || method != Z_DEFLATED ||
245
windowBits < 8 || windowBits > 15 || level < 0 || level > 9 ||
246
strategy < 0 || strategy > Z_HUFFMAN_ONLY) {
247
return Z_STREAM_ERROR;
249
s = (deflate_state *) ZALLOC(strm, 1, sizeof(deflate_state));
250
if (s == Z_NULL) return Z_MEM_ERROR;
251
strm->state = (struct internal_state FAR *)s;
254
s->noheader = noheader;
255
s->w_bits = windowBits;
256
s->w_size = 1 << s->w_bits;
257
s->w_mask = s->w_size - 1;
259
s->hash_bits = memLevel + 7;
260
s->hash_size = 1 << s->hash_bits;
261
s->hash_mask = s->hash_size - 1;
262
s->hash_shift = ((s->hash_bits+MIN_MATCH-1)/MIN_MATCH);
264
s->window = (Bytef *) ZALLOC(strm, s->w_size, 2*sizeof(Byte));
265
s->prev = (Posf *) ZALLOC(strm, s->w_size, sizeof(Pos));
266
s->head = (Posf *) ZALLOC(strm, s->hash_size, sizeof(Pos));
268
s->lit_bufsize = 1 << (memLevel + 6); /* 16K elements by default */
270
overlay = (ushf *) ZALLOC(strm, s->lit_bufsize, sizeof(ush)+2);
271
s->pending_buf = (uchf *) overlay;
272
s->pending_buf_size = (ulg)s->lit_bufsize * (sizeof(ush)+2L);
274
if (s->window == Z_NULL || s->prev == Z_NULL || s->head == Z_NULL ||
275
s->pending_buf == Z_NULL) {
276
strm->msg = (char*)ERR_MSG(Z_MEM_ERROR);
280
s->d_buf = overlay + s->lit_bufsize/sizeof(ush);
281
s->l_buf = s->pending_buf + (1+sizeof(ush))*s->lit_bufsize;
284
s->strategy = strategy;
285
s->method = (Byte)method;
287
return deflateReset(strm);
290
/* ========================================================================= */
291
int ZEXPORT deflateSetDictionary (strm, dictionary, dictLength)
293
const Bytef *dictionary;
297
uInt length = dictLength;
301
if (strm == Z_NULL || strm->state == Z_NULL || dictionary == Z_NULL ||
302
strm->state->status != INIT_STATE) return Z_STREAM_ERROR;
305
strm->adler = adler32(strm->adler, dictionary, dictLength);
307
if (length < MIN_MATCH) return Z_OK;
308
if (length > MAX_DIST(s)) {
309
length = MAX_DIST(s);
310
#ifndef USE_DICT_HEAD
311
dictionary += dictLength - length; /* use the tail of the dictionary */
314
zmemcpy(s->window, dictionary, length);
315
s->strstart = length;
316
s->block_start = (long)length;
318
/* Insert all strings in the hash table (except for the last two bytes).
319
* s->lookahead stays null, so s->ins_h will be recomputed at the next
320
* call of fill_window.
322
s->ins_h = s->window[0];
323
UPDATE_HASH(s, s->ins_h, s->window[1]);
324
for (n = 0; n <= length - MIN_MATCH; n++) {
325
INSERT_STRING(s, n, hash_head);
327
if (hash_head) hash_head = 0; /* to make compiler happy */
331
/* ========================================================================= */
332
int ZEXPORT deflateReset (strm)
337
if (strm == Z_NULL || strm->state == Z_NULL ||
338
strm->zalloc == Z_NULL || strm->zfree == Z_NULL) return Z_STREAM_ERROR;
340
strm->total_in = strm->total_out = 0;
341
strm->msg = Z_NULL; /* use zfree if we ever allocate msg dynamically */
342
strm->data_type = Z_UNKNOWN;
344
s = (deflate_state *)strm->state;
346
s->pending_out = s->pending_buf;
348
if (s->noheader < 0) {
349
s->noheader = 0; /* was set to -1 by deflate(..., Z_FINISH); */
351
s->status = s->noheader ? BUSY_STATE : INIT_STATE;
353
s->last_flush = Z_NO_FLUSH;
361
/* ========================================================================= */
362
int ZEXPORT deflateParams(strm, level, strategy)
371
if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR;
374
if (level == Z_DEFAULT_COMPRESSION) {
377
if (level < 0 || level > 9 || strategy < 0 || strategy > Z_HUFFMAN_ONLY) {
378
return Z_STREAM_ERROR;
380
func = configuration_table[s->level].func;
382
if (func != configuration_table[level].func && strm->total_in != 0) {
383
/* Flush the last buffer: */
384
err = deflate(strm, Z_PARTIAL_FLUSH);
386
if (s->level != level) {
388
s->max_lazy_match = configuration_table[level].max_lazy;
389
s->good_match = configuration_table[level].good_length;
390
s->nice_match = configuration_table[level].nice_length;
391
s->max_chain_length = configuration_table[level].max_chain;
393
s->strategy = strategy;
397
/* =========================================================================
398
* Put a short in the pending buffer. The 16-bit value is put in MSB order.
399
* IN assertion: the stream state is correct and there is enough room in
402
local void putShortMSB (s, b)
406
put_byte(s, (Byte)(b >> 8));
407
put_byte(s, (Byte)(b & 0xff));
410
/* =========================================================================
411
* Flush as much pending output as possible. All deflate() output goes
412
* through this function so some applications may wish to modify it
413
* to avoid allocating a large strm->next_out buffer and copying into it.
414
* (See also read_buf()).
416
local void flush_pending(strm)
419
unsigned len = strm->state->pending;
421
if (len > strm->avail_out) len = strm->avail_out;
422
if (len == 0) return;
424
zmemcpy(strm->next_out, strm->state->pending_out, len);
425
strm->next_out += len;
426
strm->state->pending_out += len;
427
strm->total_out += len;
428
strm->avail_out -= len;
429
strm->state->pending -= len;
430
if (strm->state->pending == 0) {
431
strm->state->pending_out = strm->state->pending_buf;
435
/* ========================================================================= */
436
int ZEXPORT deflate (strm, flush)
440
int old_flush; /* value of flush param for previous deflate call */
443
if (strm == Z_NULL || strm->state == Z_NULL ||
444
flush > Z_FINISH || flush < 0) {
445
return Z_STREAM_ERROR;
449
if (strm->next_out == Z_NULL ||
450
(strm->next_in == Z_NULL && strm->avail_in != 0) ||
451
(s->status == FINISH_STATE && flush != Z_FINISH)) {
452
ERR_RETURN(strm, Z_STREAM_ERROR);
454
if (strm->avail_out == 0) ERR_RETURN(strm, Z_BUF_ERROR);
456
s->strm = strm; /* just in case */
457
old_flush = s->last_flush;
458
s->last_flush = flush;
460
/* Write the zlib header */
461
if (s->status == INIT_STATE) {
463
uInt header = (Z_DEFLATED + ((s->w_bits-8)<<4)) << 8;
464
uInt level_flags = (s->level-1) >> 1;
466
if (level_flags > 3) level_flags = 3;
467
header |= (level_flags << 6);
468
if (s->strstart != 0) header |= PRESET_DICT;
469
header += 31 - (header % 31);
471
s->status = BUSY_STATE;
472
putShortMSB(s, header);
474
/* Save the adler32 of the preset dictionary: */
475
if (s->strstart != 0) {
476
putShortMSB(s, (uInt)(strm->adler >> 16));
477
putShortMSB(s, (uInt)(strm->adler & 0xffff));
482
/* Flush as much pending output as possible */
483
if (s->pending != 0) {
485
if (strm->avail_out == 0) {
486
/* Since avail_out is 0, deflate will be called again with
487
* more output space, but possibly with both pending and
488
* avail_in equal to zero. There won't be anything to do,
489
* but this is not an error situation so make sure we
490
* return OK instead of BUF_ERROR at next call of deflate:
496
/* Make sure there is something to do and avoid duplicate consecutive
497
* flushes. For repeated and useless calls with Z_FINISH, we keep
498
* returning Z_STREAM_END instead of Z_BUFF_ERROR.
500
} else if (strm->avail_in == 0 && flush <= old_flush &&
502
ERR_RETURN(strm, Z_BUF_ERROR);
505
/* User must not provide more input after the first FINISH: */
506
if (s->status == FINISH_STATE && strm->avail_in != 0) {
507
ERR_RETURN(strm, Z_BUF_ERROR);
510
/* Start a new block or continue the current one.
512
if (strm->avail_in != 0 || s->lookahead != 0 ||
513
(flush != Z_NO_FLUSH && s->status != FINISH_STATE)) {
516
bstate = (*(configuration_table[s->level].func))(s, flush);
518
if (bstate == finish_started || bstate == finish_done) {
519
s->status = FINISH_STATE;
521
if (bstate == need_more || bstate == finish_started) {
522
if (strm->avail_out == 0) {
523
s->last_flush = -1; /* avoid BUF_ERROR next call, see above */
526
/* If flush != Z_NO_FLUSH && avail_out == 0, the next call
527
* of deflate should use the same flush parameter to make sure
528
* that the flush is complete. So we don't have to output an
529
* empty block here, this will be done at next call. This also
530
* ensures that for a very small output buffer, we emit at most
534
if (bstate == block_done) {
535
if (flush == Z_PARTIAL_FLUSH) {
537
} else { /* FULL_FLUSH or SYNC_FLUSH */
538
_tr_stored_block(s, (char*)0, 0L, 0);
539
/* For a full flush, this empty block will be recognized
540
* as a special marker by inflate_sync().
542
if (flush == Z_FULL_FLUSH) {
543
CLEAR_HASH(s); /* forget history */
547
if (strm->avail_out == 0) {
548
s->last_flush = -1; /* avoid BUF_ERROR at next call, see above */
553
Assert(strm->avail_out > 0, "bug2");
555
if (flush != Z_FINISH) return Z_OK;
556
if (s->noheader) return Z_STREAM_END;
558
/* Write the zlib trailer (adler32) */
559
putShortMSB(s, (uInt)(strm->adler >> 16));
560
putShortMSB(s, (uInt)(strm->adler & 0xffff));
562
/* If avail_out is zero, the application will call deflate again
565
s->noheader = -1; /* write the trailer only once! */
566
return s->pending != 0 ? Z_OK : Z_STREAM_END;
569
/* ========================================================================= */
570
int ZEXPORT deflateEnd (strm)
575
if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR;
577
status = strm->state->status;
578
if (status != INIT_STATE && status != BUSY_STATE &&
579
status != FINISH_STATE) {
580
return Z_STREAM_ERROR;
583
/* Deallocate in reverse order of allocations: */
584
TRY_FREE(strm, strm->state->pending_buf);
585
TRY_FREE(strm, strm->state->head);
586
TRY_FREE(strm, strm->state->prev);
587
TRY_FREE(strm, strm->state->window);
589
ZFREE(strm, strm->state);
590
strm->state = Z_NULL;
592
return status == BUSY_STATE ? Z_DATA_ERROR : Z_OK;
595
/* =========================================================================
596
* Copy the source state to the destination state.
597
* To simplify the source, this is not supported for 16-bit MSDOS (which
598
* doesn't have enough memory anyway to duplicate compression states).
600
int ZEXPORT deflateCopy (dest, source)
605
return Z_STREAM_ERROR;
612
if (source == Z_NULL || dest == Z_NULL || source->state == Z_NULL) {
613
return Z_STREAM_ERROR;
620
ds = (deflate_state *) ZALLOC(dest, 1, sizeof(deflate_state));
621
if (ds == Z_NULL) return Z_MEM_ERROR;
622
dest->state = (struct internal_state FAR *) ds;
626
ds->window = (Bytef *) ZALLOC(dest, ds->w_size, 2*sizeof(Byte));
627
ds->prev = (Posf *) ZALLOC(dest, ds->w_size, sizeof(Pos));
628
ds->head = (Posf *) ZALLOC(dest, ds->hash_size, sizeof(Pos));
629
overlay = (ushf *) ZALLOC(dest, ds->lit_bufsize, sizeof(ush)+2);
630
ds->pending_buf = (uchf *) overlay;
632
if (ds->window == Z_NULL || ds->prev == Z_NULL || ds->head == Z_NULL ||
633
ds->pending_buf == Z_NULL) {
637
/* following zmemcpy do not work for 16-bit MSDOS */
638
zmemcpy(ds->window, ss->window, ds->w_size * 2 * sizeof(Byte));
639
zmemcpy(ds->prev, ss->prev, ds->w_size * sizeof(Pos));
640
zmemcpy(ds->head, ss->head, ds->hash_size * sizeof(Pos));
641
zmemcpy(ds->pending_buf, ss->pending_buf, (uInt)ds->pending_buf_size);
643
ds->pending_out = ds->pending_buf + (ss->pending_out - ss->pending_buf);
644
ds->d_buf = overlay + ds->lit_bufsize/sizeof(ush);
645
ds->l_buf = ds->pending_buf + (1+sizeof(ush))*ds->lit_bufsize;
647
ds->l_desc.dyn_tree = ds->dyn_ltree;
648
ds->d_desc.dyn_tree = ds->dyn_dtree;
649
ds->bl_desc.dyn_tree = ds->bl_tree;
655
/* ===========================================================================
656
* Read a new buffer from the current input stream, update the adler32
657
* and total number of bytes read. All deflate() input goes through
658
* this function so some applications may wish to modify it to avoid
659
* allocating a large strm->next_in buffer and copying from it.
660
* (See also flush_pending()).
662
local int read_buf(strm, buf, size)
667
unsigned len = strm->avail_in;
669
if (len > size) len = size;
670
if (len == 0) return 0;
672
strm->avail_in -= len;
674
if (!strm->state->noheader) {
675
strm->adler = adler32(strm->adler, strm->next_in, len);
677
zmemcpy(buf, strm->next_in, len);
678
strm->next_in += len;
679
strm->total_in += len;
684
/* ===========================================================================
685
* Initialize the "longest match" routines for a new zlib stream
687
local void lm_init (s)
690
s->window_size = (ulg)2L*s->w_size;
694
/* Set the default configuration parameters:
696
s->max_lazy_match = configuration_table[s->level].max_lazy;
697
s->good_match = configuration_table[s->level].good_length;
698
s->nice_match = configuration_table[s->level].nice_length;
699
s->max_chain_length = configuration_table[s->level].max_chain;
704
s->match_length = s->prev_length = MIN_MATCH-1;
705
s->match_available = 0;
708
match_init(); /* initialize the asm code */
712
/* ===========================================================================
713
* Set match_start to the longest match starting at the given string and
714
* return its length. Matches shorter or equal to prev_length are discarded,
715
* in which case the result is equal to prev_length and match_start is
717
* IN assertions: cur_match is the head of the hash chain for the current
718
* string (strstart) and its distance is <= MAX_DIST, and prev_length >= 1
719
* OUT assertion: the match length is not greater than s->lookahead.
722
/* For 80x86 and 680x0, an optimized version will be provided in match.asm or
723
* match.S. The code will be functionally equivalent.
726
local uInt longest_match(s, cur_match)
728
IPos cur_match; /* current match */
730
unsigned chain_length = s->max_chain_length;/* max hash chain length */
731
register Bytef *scan = s->window + s->strstart; /* current string */
732
register Bytef *match; /* matched string */
733
register int len; /* length of current match */
734
int best_len = s->prev_length; /* best match length so far */
735
int nice_match = s->nice_match; /* stop if match long enough */
736
IPos limit = s->strstart > (IPos)MAX_DIST(s) ?
737
s->strstart - (IPos)MAX_DIST(s) : NIL;
738
/* Stop when cur_match becomes <= limit. To simplify the code,
739
* we prevent matches with the string of window index 0.
741
Posf *prev = s->prev;
742
uInt wmask = s->w_mask;
745
/* Compare two bytes at a time. Note: this is not always beneficial.
746
* Try with and without -DUNALIGNED_OK to check.
748
register Bytef *strend = s->window + s->strstart + MAX_MATCH - 1;
749
register ush scan_start = *(ushf*)scan;
750
register ush scan_end = *(ushf*)(scan+best_len-1);
752
register Bytef *strend = s->window + s->strstart + MAX_MATCH;
753
register Byte scan_end1 = scan[best_len-1];
754
register Byte scan_end = scan[best_len];
757
/* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16.
758
* It is easy to get rid of this optimization if necessary.
760
Assert(s->hash_bits >= 8 && MAX_MATCH == 258, "Code too clever");
762
/* Do not waste too much time if we already have a good match: */
763
if (s->prev_length >= s->good_match) {
766
/* Do not look for matches beyond the end of the input. This is necessary
767
* to make deflate deterministic.
769
if ((uInt)nice_match > s->lookahead) nice_match = s->lookahead;
771
Assert((ulg)s->strstart <= s->window_size-MIN_LOOKAHEAD, "need lookahead");
774
Assert(cur_match < s->strstart, "no future");
775
match = s->window + cur_match;
777
/* Skip to next match if the match length cannot increase
778
* or if the match length is less than 2:
780
#if (defined(UNALIGNED_OK) && MAX_MATCH == 258)
781
/* This code assumes sizeof(unsigned short) == 2. Do not use
782
* UNALIGNED_OK if your compiler uses a different size.
784
if (*(ushf*)(match+best_len-1) != scan_end ||
785
*(ushf*)match != scan_start) continue;
787
/* It is not necessary to compare scan[2] and match[2] since they are
788
* always equal when the other bytes match, given that the hash keys
789
* are equal and that HASH_BITS >= 8. Compare 2 bytes at a time at
790
* strstart+3, +5, ... up to strstart+257. We check for insufficient
791
* lookahead only every 4th comparison; the 128th check will be made
792
* at strstart+257. If MAX_MATCH-2 is not a multiple of 8, it is
793
* necessary to put more guard bytes at the end of the window, or
794
* to check more often for insufficient lookahead.
796
Assert(scan[2] == match[2], "scan[2]?");
799
} while (*(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
800
*(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
801
*(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
802
*(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
804
/* The funny "do {}" generates better code on most compilers */
806
/* Here, scan <= window+strstart+257 */
807
Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan");
808
if (*scan == *match) scan++;
810
len = (MAX_MATCH - 1) - (int)(strend-scan);
811
scan = strend - (MAX_MATCH-1);
813
#else /* UNALIGNED_OK */
815
if (match[best_len] != scan_end ||
816
match[best_len-1] != scan_end1 ||
818
*++match != scan[1]) continue;
820
/* The check at best_len-1 can be removed because it will be made
821
* again later. (This heuristic is not always a win.)
822
* It is not necessary to compare scan[2] and match[2] since they
823
* are always equal when the other bytes match, given that
824
* the hash keys are equal and that HASH_BITS >= 8.
827
Assert(*scan == *match, "match[2]?");
829
/* We check for insufficient lookahead only every 8th comparison;
830
* the 256th check will be made at strstart+258.
833
} while (*++scan == *++match && *++scan == *++match &&
834
*++scan == *++match && *++scan == *++match &&
835
*++scan == *++match && *++scan == *++match &&
836
*++scan == *++match && *++scan == *++match &&
839
Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan");
841
len = MAX_MATCH - (int)(strend - scan);
842
scan = strend - MAX_MATCH;
844
#endif /* UNALIGNED_OK */
846
if (len > best_len) {
847
s->match_start = cur_match;
849
if (len >= nice_match) break;
851
scan_end = *(ushf*)(scan+best_len-1);
853
scan_end1 = scan[best_len-1];
854
scan_end = scan[best_len];
857
} while ((cur_match = prev[cur_match & wmask]) > limit
858
&& --chain_length != 0);
860
if ((uInt)best_len <= s->lookahead) return (uInt)best_len;
865
/* ---------------------------------------------------------------------------
866
* Optimized version for level == 1 only
868
local uInt longest_match(s, cur_match)
870
IPos cur_match; /* current match */
872
register Bytef *scan = s->window + s->strstart; /* current string */
873
register Bytef *match; /* matched string */
874
register int len; /* length of current match */
875
register Bytef *strend = s->window + s->strstart + MAX_MATCH;
877
/* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16.
878
* It is easy to get rid of this optimization if necessary.
880
Assert(s->hash_bits >= 8 && MAX_MATCH == 258, "Code too clever");
882
Assert((ulg)s->strstart <= s->window_size-MIN_LOOKAHEAD, "need lookahead");
884
Assert(cur_match < s->strstart, "no future");
886
match = s->window + cur_match;
888
/* Return failure if the match length is less than 2:
890
if (match[0] != scan[0] || match[1] != scan[1]) return MIN_MATCH-1;
892
/* The check at best_len-1 can be removed because it will be made
893
* again later. (This heuristic is not always a win.)
894
* It is not necessary to compare scan[2] and match[2] since they
895
* are always equal when the other bytes match, given that
896
* the hash keys are equal and that HASH_BITS >= 8.
898
scan += 2, match += 2;
899
Assert(*scan == *match, "match[2]?");
901
/* We check for insufficient lookahead only every 8th comparison;
902
* the 256th check will be made at strstart+258.
905
} while (*++scan == *++match && *++scan == *++match &&
906
*++scan == *++match && *++scan == *++match &&
907
*++scan == *++match && *++scan == *++match &&
908
*++scan == *++match && *++scan == *++match &&
911
Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan");
913
len = MAX_MATCH - (int)(strend - scan);
915
if (len < MIN_MATCH) return MIN_MATCH - 1;
917
s->match_start = cur_match;
918
return len <= s->lookahead ? len : s->lookahead;
924
/* ===========================================================================
925
* Check that the match at match_start is indeed a match.
927
local void check_match(s, start, match, length)
932
/* check that the match is indeed a match */
933
if (zmemcmp(s->window + match,
934
s->window + start, length) != EQUAL) {
935
fprintf(stderr, " start %u, match %u, length %d\n",
936
start, match, length);
938
fprintf(stderr, "%c%c", s->window[match++], s->window[start++]);
939
} while (--length != 0);
940
z_error("invalid match");
943
fprintf(stderr,"\\[%d,%d]", start-match, length);
944
do { putc(s->window[start++], stderr); } while (--length != 0);
948
# define check_match(s, start, match, length)
951
/* ===========================================================================
952
* Fill the window when the lookahead becomes insufficient.
953
* Updates strstart and lookahead.
955
* IN assertion: lookahead < MIN_LOOKAHEAD
956
* OUT assertions: strstart <= window_size-MIN_LOOKAHEAD
957
* At least one byte has been read, or avail_in == 0; reads are
958
* performed for at least two bytes (required for the zip translate_eol
959
* option -- not supported here).
961
local void fill_window(s)
964
register unsigned n, m;
966
unsigned more; /* Amount of free space at the end of the window. */
967
uInt wsize = s->w_size;
970
more = (unsigned)(s->window_size -(ulg)s->lookahead -(ulg)s->strstart);
972
/* Deal with !@#$% 64K limit: */
973
if (more == 0 && s->strstart == 0 && s->lookahead == 0) {
976
} else if (more == (unsigned)(-1)) {
977
/* Very unlikely, but possible on 16 bit machine if strstart == 0
978
* and lookahead == 1 (input done one byte at time)
982
/* If the window is almost full and there is insufficient lookahead,
983
* move the upper half to the lower one to make room in the upper half.
985
} else if (s->strstart >= wsize+MAX_DIST(s)) {
987
zmemcpy(s->window, s->window+wsize, (unsigned)wsize);
988
s->match_start -= wsize;
989
s->strstart -= wsize; /* we now have strstart >= MAX_DIST */
990
s->block_start -= (long) wsize;
992
/* Slide the hash table (could be avoided with 32 bit values
993
at the expense of memory usage). We slide even when level == 0
994
to keep the hash table consistent if we switch back to level > 0
995
later. (Using level 0 permanently is not an optimal usage of
996
zlib, so we don't care about this pathological case.)
1002
*p = (Pos)(m >= wsize ? m-wsize : NIL);
1010
*p = (Pos)(m >= wsize ? m-wsize : NIL);
1011
/* If n is not on any hash chain, prev[n] is garbage but
1012
* its value will never be used.
1018
if (s->strm->avail_in == 0) return;
1020
/* If there was no sliding:
1021
* strstart <= WSIZE+MAX_DIST-1 && lookahead <= MIN_LOOKAHEAD - 1 &&
1022
* more == window_size - lookahead - strstart
1023
* => more >= window_size - (MIN_LOOKAHEAD-1 + WSIZE + MAX_DIST-1)
1024
* => more >= window_size - 2*WSIZE + 2
1025
* In the BIG_MEM or MMAP case (not yet supported),
1026
* window_size == input_size + MIN_LOOKAHEAD &&
1027
* strstart + s->lookahead <= input_size => more >= MIN_LOOKAHEAD.
1028
* Otherwise, window_size == 2*WSIZE so more >= 2.
1029
* If there was sliding, more >= WSIZE. So in all cases, more >= 2.
1031
Assert(more >= 2, "more < 2");
1033
n = read_buf(s->strm, s->window + s->strstart + s->lookahead, more);
1036
/* Initialize the hash value now that we have some input: */
1037
if (s->lookahead >= MIN_MATCH) {
1038
s->ins_h = s->window[s->strstart];
1039
UPDATE_HASH(s, s->ins_h, s->window[s->strstart+1]);
1041
Call UPDATE_HASH() MIN_MATCH-3 more times
1044
/* If the whole input has less than MIN_MATCH bytes, ins_h is garbage,
1045
* but this is not important since only literal bytes will be emitted.
1048
} while (s->lookahead < MIN_LOOKAHEAD && s->strm->avail_in != 0);
1051
/* ===========================================================================
1052
* Flush the current block, with given end-of-file flag.
1053
* IN assertion: strstart is set to the end of the current match.
1055
#define FLUSH_BLOCK_ONLY(s, eof) { \
1056
_tr_flush_block(s, (s->block_start >= 0L ? \
1057
(charf *)&s->window[(unsigned)s->block_start] : \
1059
(ulg)((long)s->strstart - s->block_start), \
1061
s->block_start = s->strstart; \
1062
flush_pending(s->strm); \
1063
Tracev((stderr,"[FLUSH]")); \
1066
/* Same but force premature exit if necessary. */
1067
#define FLUSH_BLOCK(s, eof) { \
1068
FLUSH_BLOCK_ONLY(s, eof); \
1069
if (s->strm->avail_out == 0) return (eof) ? finish_started : need_more; \
1072
/* ===========================================================================
1073
* Copy without compression as much as possible from the input stream, return
1074
* the current block state.
1075
* This function does not insert new strings in the dictionary since
1076
* uncompressible data is probably not useful. This function is used
1077
* only for the level=0 compression option.
1078
* NOTE: this function should be optimized to avoid extra copying from
1079
* window to pending_buf.
1081
local block_state deflate_stored(s, flush)
1085
/* Stored blocks are limited to 0xffff bytes, pending_buf is limited
1086
* to pending_buf_size, and each stored block has a 5 byte header:
1088
ulg max_block_size = 0xffff;
1091
if (max_block_size > s->pending_buf_size - 5) {
1092
max_block_size = s->pending_buf_size - 5;
1095
/* Copy as much as possible from input to output: */
1097
/* Fill the window as much as possible: */
1098
if (s->lookahead <= 1) {
1100
Assert(s->strstart < s->w_size+MAX_DIST(s) ||
1101
s->block_start >= (long)s->w_size, "slide too late");
1104
if (s->lookahead == 0 && flush == Z_NO_FLUSH) return need_more;
1106
if (s->lookahead == 0) break; /* flush the current block */
1108
Assert(s->block_start >= 0L, "block gone");
1110
s->strstart += s->lookahead;
1113
/* Emit a stored block if pending_buf will be full: */
1114
max_start = s->block_start + max_block_size;
1115
if (s->strstart == 0 || (ulg)s->strstart >= max_start) {
1116
/* strstart == 0 is possible when wraparound on 16-bit machine */
1117
s->lookahead = (uInt)(s->strstart - max_start);
1118
s->strstart = (uInt)max_start;
1121
/* Flush if we may have to slide, otherwise block_start may become
1122
* negative and the data will be gone:
1124
if (s->strstart - (uInt)s->block_start >= MAX_DIST(s)) {
1128
FLUSH_BLOCK(s, flush == Z_FINISH);
1129
return flush == Z_FINISH ? finish_done : block_done;
1132
/* ===========================================================================
1133
* Compress as much as possible from the input stream, return the current
1135
* This function does not perform lazy evaluation of matches and inserts
1136
* new strings in the dictionary only for unmatched strings or for short
1137
* matches. It is used only for the fast compression options.
1139
local block_state deflate_fast(s, flush)
1143
IPos hash_head = NIL; /* head of the hash chain */
1144
int bflush; /* set if current block must be flushed */
1147
/* Make sure that we always have enough lookahead, except
1148
* at the end of the input file. We need MAX_MATCH bytes
1149
* for the next match, plus MIN_MATCH bytes to insert the
1150
* string following the next match.
1152
if (s->lookahead < MIN_LOOKAHEAD) {
1154
if (s->lookahead < MIN_LOOKAHEAD && flush == Z_NO_FLUSH) {
1157
if (s->lookahead == 0) break; /* flush the current block */
1160
/* Insert the string window[strstart .. strstart+2] in the
1161
* dictionary, and set hash_head to the head of the hash chain:
1163
if (s->lookahead >= MIN_MATCH) {
1164
INSERT_STRING(s, s->strstart, hash_head);
1167
/* Find the longest match, discarding those <= prev_length.
1168
* At this point we have always match_length < MIN_MATCH
1170
if (hash_head != NIL && s->strstart - hash_head <= MAX_DIST(s)) {
1171
/* To simplify the code, we prevent matches with the string
1172
* of window index 0 (in particular we have to avoid a match
1173
* of the string with itself at the start of the input file).
1175
if (s->strategy != Z_HUFFMAN_ONLY) {
1176
s->match_length = longest_match (s, hash_head);
1178
/* longest_match() sets match_start */
1180
if (s->match_length >= MIN_MATCH) {
1181
check_match(s, s->strstart, s->match_start, s->match_length);
1183
_tr_tally_dist(s, s->strstart - s->match_start,
1184
s->match_length - MIN_MATCH, bflush);
1186
s->lookahead -= s->match_length;
1188
/* Insert new strings in the hash table only if the match length
1189
* is not too large. This saves time but degrades compression.
1192
if (s->match_length <= s->max_insert_length &&
1193
s->lookahead >= MIN_MATCH) {
1194
s->match_length--; /* string at strstart already in hash table */
1197
INSERT_STRING(s, s->strstart, hash_head);
1198
/* strstart never exceeds WSIZE-MAX_MATCH, so there are
1199
* always MIN_MATCH bytes ahead.
1201
} while (--s->match_length != 0);
1206
s->strstart += s->match_length;
1207
s->match_length = 0;
1208
s->ins_h = s->window[s->strstart];
1209
UPDATE_HASH(s, s->ins_h, s->window[s->strstart+1]);
1211
Call UPDATE_HASH() MIN_MATCH-3 more times
1213
/* If lookahead < MIN_MATCH, ins_h is garbage, but it does not
1214
* matter since it will be recomputed at next deflate call.
1218
/* No match, output a literal byte */
1219
Tracevv((stderr,"%c", s->window[s->strstart]));
1220
_tr_tally_lit (s, s->window[s->strstart], bflush);
1224
if (bflush) FLUSH_BLOCK(s, 0);
1226
FLUSH_BLOCK(s, flush == Z_FINISH);
1227
return flush == Z_FINISH ? finish_done : block_done;
1230
/* ===========================================================================
1231
* Same as above, but achieves better compression. We use a lazy
1232
* evaluation for matches: a match is finally adopted only if there is
1233
* no better match at the next window position.
1235
local block_state deflate_slow(s, flush)
1239
IPos hash_head = NIL; /* head of hash chain */
1240
int bflush; /* set if current block must be flushed */
1242
/* Process the input block. */
1244
/* Make sure that we always have enough lookahead, except
1245
* at the end of the input file. We need MAX_MATCH bytes
1246
* for the next match, plus MIN_MATCH bytes to insert the
1247
* string following the next match.
1249
if (s->lookahead < MIN_LOOKAHEAD) {
1251
if (s->lookahead < MIN_LOOKAHEAD && flush == Z_NO_FLUSH) {
1254
if (s->lookahead == 0) break; /* flush the current block */
1257
/* Insert the string window[strstart .. strstart+2] in the
1258
* dictionary, and set hash_head to the head of the hash chain:
1260
if (s->lookahead >= MIN_MATCH) {
1261
INSERT_STRING(s, s->strstart, hash_head);
1264
/* Find the longest match, discarding those <= prev_length.
1266
s->prev_length = s->match_length, s->prev_match = s->match_start;
1267
s->match_length = MIN_MATCH-1;
1269
if (hash_head != NIL && s->prev_length < s->max_lazy_match &&
1270
s->strstart - hash_head <= MAX_DIST(s)) {
1271
/* To simplify the code, we prevent matches with the string
1272
* of window index 0 (in particular we have to avoid a match
1273
* of the string with itself at the start of the input file).
1275
if (s->strategy != Z_HUFFMAN_ONLY) {
1276
s->match_length = longest_match (s, hash_head);
1278
/* longest_match() sets match_start */
1280
if (s->match_length <= 5 && (s->strategy == Z_FILTERED ||
1281
(s->match_length == MIN_MATCH &&
1282
s->strstart - s->match_start > TOO_FAR))) {
1284
/* If prev_match is also MIN_MATCH, match_start is garbage
1285
* but we will ignore the current match anyway.
1287
s->match_length = MIN_MATCH-1;
1290
/* If there was a match at the previous step and the current
1291
* match is not better, output the previous match:
1293
if (s->prev_length >= MIN_MATCH && s->match_length <= s->prev_length) {
1294
uInt max_insert = s->strstart + s->lookahead - MIN_MATCH;
1295
/* Do not insert strings in hash table beyond this. */
1297
check_match(s, s->strstart-1, s->prev_match, s->prev_length);
1299
_tr_tally_dist(s, s->strstart -1 - s->prev_match,
1300
s->prev_length - MIN_MATCH, bflush);
1302
/* Insert in hash table all strings up to the end of the match.
1303
* strstart-1 and strstart are already inserted. If there is not
1304
* enough lookahead, the last two strings are not inserted in
1307
s->lookahead -= s->prev_length-1;
1308
s->prev_length -= 2;
1310
if (++s->strstart <= max_insert) {
1311
INSERT_STRING(s, s->strstart, hash_head);
1313
} while (--s->prev_length != 0);
1314
s->match_available = 0;
1315
s->match_length = MIN_MATCH-1;
1318
if (bflush) FLUSH_BLOCK(s, 0);
1320
} else if (s->match_available) {
1321
/* If there was no match at the previous position, output a
1322
* single literal. If there was a match but the current match
1323
* is longer, truncate the previous match to a single literal.
1325
Tracevv((stderr,"%c", s->window[s->strstart-1]));
1326
_tr_tally_lit(s, s->window[s->strstart-1], bflush);
1328
FLUSH_BLOCK_ONLY(s, 0);
1332
if (s->strm->avail_out == 0) return need_more;
1334
/* There is no previous match to compare with, wait for
1335
* the next step to decide.
1337
s->match_available = 1;
1342
Assert (flush != Z_NO_FLUSH, "no flush?");
1343
if (s->match_available) {
1344
Tracevv((stderr,"%c", s->window[s->strstart-1]));
1345
_tr_tally_lit(s, s->window[s->strstart-1], bflush);
1346
s->match_available = 0;
1348
FLUSH_BLOCK(s, flush == Z_FINISH);
1349
return flush == Z_FINISH ? finish_done : block_done;
added
removed
gdchart0.94c/zlib-1.1.3/deflate.c
