1
/* deflate.c -- compress data using the deflation algorithm
2
* Copyright (C) 1995-2003 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 http://www.ietf.org/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 2004/12/24 20:46:30 kakaroto Exp $ */
54
const char deflate_copyright[] =
55
" deflate 1.2.1 Copyright 1995-2003 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));
80
local block_state deflate_slow OF((deflate_state *s, int flush));
82
local void lm_init OF((deflate_state *s));
83
local void putShortMSB OF((deflate_state *s, uInt b));
84
local void flush_pending OF((z_streamp strm));
85
local int read_buf OF((z_streamp strm, Bytef *buf, unsigned size));
88
void match_init OF((void)); /* asm code initialization */
89
uInt longest_match OF((deflate_state *s, IPos cur_match));
91
local uInt longest_match OF((deflate_state *s, IPos cur_match));
94
local uInt longest_match_fast OF((deflate_state *s, IPos cur_match));
97
local void check_match OF((deflate_state *s, IPos start, IPos match,
101
/* ===========================================================================
106
/* Tail of hash chains */
109
# define TOO_FAR 4096
111
/* Matches of length 3 are discarded if their distance exceeds TOO_FAR */
113
#define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1)
114
/* Minimum amount of lookahead, except at the end of the input file.
115
* See deflate.c for comments about the MIN_MATCH+1.
118
/* Values for max_lazy_match, good_match and max_chain_length, depending on
119
* the desired pack level (0..9). The values given below have been tuned to
120
* exclude worst case performance for pathological files. Better values may be
121
* found for specific files.
123
typedef struct config_s {
124
ush good_length; /* reduce lazy search above this match length */
125
ush max_lazy; /* do not perform lazy search above this match length */
126
ush nice_length; /* quit search above this match length */
132
local const config configuration_table[2] = {
133
/* good lazy nice chain */
134
/* 0 */ {0, 0, 0, 0, deflate_stored}, /* store only */
135
/* 1 */ {4, 4, 8, 4, deflate_fast}}; /* max speed, no lazy matches */
137
local const config configuration_table[10] = {
138
/* good lazy nice chain */
139
/* 0 */ {0, 0, 0, 0, deflate_stored}, /* store only */
140
/* 1 */ {4, 4, 8, 4, deflate_fast}, /* max speed, no lazy matches */
141
/* 2 */ {4, 5, 16, 8, deflate_fast},
142
/* 3 */ {4, 6, 32, 32, deflate_fast},
144
/* 4 */ {4, 4, 16, 16, deflate_slow}, /* lazy matches */
145
/* 5 */ {8, 16, 32, 32, deflate_slow},
146
/* 6 */ {8, 16, 128, 128, deflate_slow},
147
/* 7 */ {8, 32, 128, 256, deflate_slow},
148
/* 8 */ {32, 128, 258, 1024, deflate_slow},
149
/* 9 */ {32, 258, 258, 4096, deflate_slow}}; /* max compression */
152
/* Note: the deflate() code requires max_lazy >= MIN_MATCH and max_chain >= 4
153
* For deflate_fast() (levels <= 3) good is ignored and lazy has a different
158
/* result of memcmp for equal strings */
160
#ifndef NO_DUMMY_DECL
161
struct static_tree_desc_s {int dummy;}; /* for buggy compilers */
164
/* ===========================================================================
165
* Update a hash value with the given input byte
166
* IN assertion: all calls to to UPDATE_HASH are made with consecutive
167
* input characters, so that a running hash key can be computed from the
168
* previous key instead of complete recalculation each time.
170
#define UPDATE_HASH(s,h,c) (h = (((h)<<s->hash_shift) ^ (c)) & s->hash_mask)
173
/* ===========================================================================
174
* Insert string str in the dictionary and set match_head to the previous head
175
* of the hash chain (the most recent string with same hash key). Return
176
* the previous length of the hash chain.
177
* If this file is compiled with -DFASTEST, the compression level is forced
178
* to 1, and no hash chains are maintained.
179
* IN assertion: all calls to to INSERT_STRING are made with consecutive
180
* input characters and the first MIN_MATCH bytes of str are valid
181
* (except for the last MIN_MATCH-1 bytes of the input file).
184
#define INSERT_STRING(s, str, match_head) \
185
(UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \
186
match_head = s->head[s->ins_h], \
187
s->head[s->ins_h] = (Pos)(str))
189
#define INSERT_STRING(s, str, match_head) \
190
(UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \
191
match_head = s->prev[(str) & s->w_mask] = s->head[s->ins_h], \
192
s->head[s->ins_h] = (Pos)(str))
195
/* ===========================================================================
196
* Initialize the hash table (avoiding 64K overflow for 16 bit systems).
197
* prev[] will be initialized on the fly.
199
#define CLEAR_HASH(s) \
200
s->head[s->hash_size-1] = NIL; \
201
zmemzero((Bytef *)s->head, (unsigned)(s->hash_size-1)*sizeof(*s->head));
203
/* ========================================================================= */
204
int ZEXPORT deflateInit_(strm, level, version, stream_size)
210
return deflateInit2_(strm, level, Z_DEFLATED, MAX_WBITS, DEF_MEM_LEVEL,
211
Z_DEFAULT_STRATEGY, version, stream_size);
212
/* To do: ignore strm->next_in if we use it as window */
215
/* ========================================================================= */
216
int ZEXPORT deflateInit2_(strm, level, method, windowBits, memLevel, strategy,
217
version, stream_size)
229
static const char my_version[] = ZLIB_VERSION;
232
/* We overlay pending_buf and d_buf+l_buf. This works since the average
233
* output size for (length,distance) codes is <= 24 bits.
236
if (version == Z_NULL || version[0] != my_version[0] ||
237
stream_size != sizeof(z_stream)) {
238
return Z_VERSION_ERROR;
240
if (strm == Z_NULL) return Z_STREAM_ERROR;
243
if (strm->zalloc == (alloc_func)0) {
244
strm->zalloc = zcalloc;
245
strm->opaque = (voidpf)0;
247
if (strm->zfree == (free_func)0) strm->zfree = zcfree;
250
if (level != 0) level = 1;
252
if (level == Z_DEFAULT_COMPRESSION) level = 6;
255
if (windowBits < 0) { /* suppress zlib wrapper */
257
windowBits = -windowBits;
260
else if (windowBits > 15) {
261
wrap = 2; /* write gzip wrapper instead */
265
if (memLevel < 1 || memLevel > MAX_MEM_LEVEL || method != Z_DEFLATED ||
266
windowBits < 8 || windowBits > 15 || level < 0 || level > 9 ||
267
strategy < 0 || strategy > Z_RLE) {
268
return Z_STREAM_ERROR;
270
if (windowBits == 8) windowBits = 9; /* until 256-byte window bug fixed */
271
s = (deflate_state *) ZALLOC(strm, 1, sizeof(deflate_state));
272
if (s == Z_NULL) return Z_MEM_ERROR;
273
strm->state = (struct internal_state FAR *)s;
277
s->w_bits = windowBits;
278
s->w_size = 1 << s->w_bits;
279
s->w_mask = s->w_size - 1;
281
s->hash_bits = memLevel + 7;
282
s->hash_size = 1 << s->hash_bits;
283
s->hash_mask = s->hash_size - 1;
284
s->hash_shift = ((s->hash_bits+MIN_MATCH-1)/MIN_MATCH);
286
s->window = (Bytef *) ZALLOC(strm, s->w_size, 2*sizeof(Byte));
287
s->prev = (Posf *) ZALLOC(strm, s->w_size, sizeof(Pos));
288
s->head = (Posf *) ZALLOC(strm, s->hash_size, sizeof(Pos));
290
s->lit_bufsize = 1 << (memLevel + 6); /* 16K elements by default */
292
overlay = (ushf *) ZALLOC(strm, s->lit_bufsize, sizeof(ush)+2);
293
s->pending_buf = (uchf *) overlay;
294
s->pending_buf_size = (ulg)s->lit_bufsize * (sizeof(ush)+2L);
296
if (s->window == Z_NULL || s->prev == Z_NULL || s->head == Z_NULL ||
297
s->pending_buf == Z_NULL) {
298
s->status = FINISH_STATE;
299
strm->msg = (char*)ERR_MSG(Z_MEM_ERROR);
303
s->d_buf = overlay + s->lit_bufsize/sizeof(ush);
304
s->l_buf = s->pending_buf + (1+sizeof(ush))*s->lit_bufsize;
307
s->strategy = strategy;
308
s->method = (Byte)method;
310
return deflateReset(strm);
313
/* ========================================================================= */
314
int ZEXPORT deflateSetDictionary (strm, dictionary, dictLength)
316
const Bytef *dictionary;
320
uInt length = dictLength;
324
if (strm == Z_NULL || strm->state == Z_NULL || dictionary == Z_NULL ||
325
strm->state->wrap == 2 ||
326
(strm->state->wrap == 1 && strm->state->status != INIT_STATE))
327
return Z_STREAM_ERROR;
331
strm->adler = adler32(strm->adler, dictionary, dictLength);
333
if (length < MIN_MATCH) return Z_OK;
334
if (length > MAX_DIST(s)) {
335
length = MAX_DIST(s);
336
#ifndef USE_DICT_HEAD
337
dictionary += dictLength - length; /* use the tail of the dictionary */
340
zmemcpy(s->window, dictionary, length);
341
s->strstart = length;
342
s->block_start = (long)length;
344
/* Insert all strings in the hash table (except for the last two bytes).
345
* s->lookahead stays null, so s->ins_h will be recomputed at the next
346
* call of fill_window.
348
s->ins_h = s->window[0];
349
UPDATE_HASH(s, s->ins_h, s->window[1]);
350
for (n = 0; n <= length - MIN_MATCH; n++) {
351
INSERT_STRING(s, n, hash_head);
353
if (hash_head) hash_head = 0; /* to make compiler happy */
357
/* ========================================================================= */
358
int ZEXPORT deflateReset (strm)
363
if (strm == Z_NULL || strm->state == Z_NULL ||
364
strm->zalloc == (alloc_func)0 || strm->zfree == (free_func)0) {
365
return Z_STREAM_ERROR;
368
strm->total_in = strm->total_out = 0;
369
strm->msg = Z_NULL; /* use zfree if we ever allocate msg dynamically */
370
strm->data_type = Z_UNKNOWN;
372
s = (deflate_state *)strm->state;
374
s->pending_out = s->pending_buf;
377
s->wrap = -s->wrap; /* was made negative by deflate(..., Z_FINISH); */
379
s->status = s->wrap ? INIT_STATE : BUSY_STATE;
382
s->wrap == 2 ? crc32(0L, Z_NULL, 0) :
384
adler32(0L, Z_NULL, 0);
385
s->last_flush = Z_NO_FLUSH;
393
/* ========================================================================= */
394
int ZEXPORT deflatePrime (strm, bits, value)
399
if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR;
400
strm->state->bi_valid = bits;
401
strm->state->bi_buf = (ush)(value & ((1 << bits) - 1));
405
/* ========================================================================= */
406
int ZEXPORT deflateParams(strm, level, strategy)
415
if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR;
419
if (level != 0) level = 1;
421
if (level == Z_DEFAULT_COMPRESSION) level = 6;
423
if (level < 0 || level > 9 || strategy < 0 || strategy > Z_RLE) {
424
return Z_STREAM_ERROR;
426
func = configuration_table[s->level].func;
428
if (func != configuration_table[level].func && strm->total_in != 0) {
429
/* Flush the last buffer: */
430
err = deflate(strm, Z_PARTIAL_FLUSH);
432
if (s->level != level) {
434
s->max_lazy_match = configuration_table[level].max_lazy;
435
s->good_match = configuration_table[level].good_length;
436
s->nice_match = configuration_table[level].nice_length;
437
s->max_chain_length = configuration_table[level].max_chain;
439
s->strategy = strategy;
443
/* =========================================================================
444
* For the default windowBits of 15 and memLevel of 8, this function returns
445
* a close to exact, as well as small, upper bound on the compressed size.
446
* They are coded as constants here for a reason--if the #define's are
447
* changed, then this function needs to be changed as well. The return
448
* value for 15 and 8 only works for those exact settings.
450
* For any setting other than those defaults for windowBits and memLevel,
451
* the value returned is a conservative worst case for the maximum expansion
452
* resulting from using fixed blocks instead of stored blocks, which deflate
453
* can emit on compressed data for some combinations of the parameters.
455
* This function could be more sophisticated to provide closer upper bounds
456
* for every combination of windowBits and memLevel, as well as wrap.
457
* But even the conservative upper bound of about 14% expansion does not
458
* seem onerous for output buffer allocation.
460
uLong ZEXPORT deflateBound(strm, sourceLen)
467
/* conservative upper bound */
468
destLen = sourceLen +
469
((sourceLen + 7) >> 3) + ((sourceLen + 63) >> 6) + 11;
471
/* if can't get parameters, return conservative bound */
472
if (strm == Z_NULL || strm->state == Z_NULL)
475
/* if not default parameters, return conservative bound */
477
if (s->w_bits != 15 || s->hash_bits != 8 + 7)
480
/* default settings: return tight bound for that case */
481
return compressBound(sourceLen);
484
/* =========================================================================
485
* Put a short in the pending buffer. The 16-bit value is put in MSB order.
486
* IN assertion: the stream state is correct and there is enough room in
489
local void putShortMSB (s, b)
493
put_byte(s, (Byte)(b >> 8));
494
put_byte(s, (Byte)(b & 0xff));
497
/* =========================================================================
498
* Flush as much pending output as possible. All deflate() output goes
499
* through this function so some applications may wish to modify it
500
* to avoid allocating a large strm->next_out buffer and copying into it.
501
* (See also read_buf()).
503
local void flush_pending(strm)
506
unsigned len = strm->state->pending;
508
if (len > strm->avail_out) len = strm->avail_out;
509
if (len == 0) return;
511
zmemcpy(strm->next_out, strm->state->pending_out, len);
512
strm->next_out += len;
513
strm->state->pending_out += len;
514
strm->total_out += len;
515
strm->avail_out -= len;
516
strm->state->pending -= len;
517
if (strm->state->pending == 0) {
518
strm->state->pending_out = strm->state->pending_buf;
522
/* ========================================================================= */
523
int ZEXPORT deflate (strm, flush)
527
int old_flush; /* value of flush param for previous deflate call */
530
if (strm == Z_NULL || strm->state == Z_NULL ||
531
flush > Z_FINISH || flush < 0) {
532
return Z_STREAM_ERROR;
536
if (strm->next_out == Z_NULL ||
537
(strm->next_in == Z_NULL && strm->avail_in != 0) ||
538
(s->status == FINISH_STATE && flush != Z_FINISH)) {
539
ERR_RETURN(strm, Z_STREAM_ERROR);
541
if (strm->avail_out == 0) ERR_RETURN(strm, Z_BUF_ERROR);
543
s->strm = strm; /* just in case */
544
old_flush = s->last_flush;
545
s->last_flush = flush;
547
/* Write the header */
548
if (s->status == INIT_STATE) {
559
put_byte(s, s->level == 9 ? 2 :
560
(s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ?
563
s->status = BUSY_STATE;
564
strm->adler = crc32(0L, Z_NULL, 0);
569
uInt header = (Z_DEFLATED + ((s->w_bits-8)<<4)) << 8;
572
if (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2)
574
else if (s->level < 6)
576
else if (s->level == 6)
580
header |= (level_flags << 6);
581
if (s->strstart != 0) header |= PRESET_DICT;
582
header += 31 - (header % 31);
584
s->status = BUSY_STATE;
585
putShortMSB(s, header);
587
/* Save the adler32 of the preset dictionary: */
588
if (s->strstart != 0) {
589
putShortMSB(s, (uInt)(strm->adler >> 16));
590
putShortMSB(s, (uInt)(strm->adler & 0xffff));
592
strm->adler = adler32(0L, Z_NULL, 0);
596
/* Flush as much pending output as possible */
597
if (s->pending != 0) {
599
if (strm->avail_out == 0) {
600
/* Since avail_out is 0, deflate will be called again with
601
* more output space, but possibly with both pending and
602
* avail_in equal to zero. There won't be anything to do,
603
* but this is not an error situation so make sure we
604
* return OK instead of BUF_ERROR at next call of deflate:
610
/* Make sure there is something to do and avoid duplicate consecutive
611
* flushes. For repeated and useless calls with Z_FINISH, we keep
612
* returning Z_STREAM_END instead of Z_BUF_ERROR.
614
} else if (strm->avail_in == 0 && flush <= old_flush &&
616
ERR_RETURN(strm, Z_BUF_ERROR);
619
/* User must not provide more input after the first FINISH: */
620
if (s->status == FINISH_STATE && strm->avail_in != 0) {
621
ERR_RETURN(strm, Z_BUF_ERROR);
624
/* Start a new block or continue the current one.
626
if (strm->avail_in != 0 || s->lookahead != 0 ||
627
(flush != Z_NO_FLUSH && s->status != FINISH_STATE)) {
630
bstate = (*(configuration_table[s->level].func))(s, flush);
632
if (bstate == finish_started || bstate == finish_done) {
633
s->status = FINISH_STATE;
635
if (bstate == need_more || bstate == finish_started) {
636
if (strm->avail_out == 0) {
637
s->last_flush = -1; /* avoid BUF_ERROR next call, see above */
640
/* If flush != Z_NO_FLUSH && avail_out == 0, the next call
641
* of deflate should use the same flush parameter to make sure
642
* that the flush is complete. So we don't have to output an
643
* empty block here, this will be done at next call. This also
644
* ensures that for a very small output buffer, we emit at most
648
if (bstate == block_done) {
649
if (flush == Z_PARTIAL_FLUSH) {
651
} else { /* FULL_FLUSH or SYNC_FLUSH */
652
_tr_stored_block(s, (char*)0, 0L, 0);
653
/* For a full flush, this empty block will be recognized
654
* as a special marker by inflate_sync().
656
if (flush == Z_FULL_FLUSH) {
657
CLEAR_HASH(s); /* forget history */
661
if (strm->avail_out == 0) {
662
s->last_flush = -1; /* avoid BUF_ERROR at next call, see above */
667
Assert(strm->avail_out > 0, "bug2");
669
if (flush != Z_FINISH) return Z_OK;
670
if (s->wrap <= 0) return Z_STREAM_END;
672
/* Write the trailer */
675
put_byte(s, (Byte)(strm->adler & 0xff));
676
put_byte(s, (Byte)((strm->adler >> 8) & 0xff));
677
put_byte(s, (Byte)((strm->adler >> 16) & 0xff));
678
put_byte(s, (Byte)((strm->adler >> 24) & 0xff));
679
put_byte(s, (Byte)(strm->total_in & 0xff));
680
put_byte(s, (Byte)((strm->total_in >> 8) & 0xff));
681
put_byte(s, (Byte)((strm->total_in >> 16) & 0xff));
682
put_byte(s, (Byte)((strm->total_in >> 24) & 0xff));
687
putShortMSB(s, (uInt)(strm->adler >> 16));
688
putShortMSB(s, (uInt)(strm->adler & 0xffff));
691
/* If avail_out is zero, the application will call deflate again
694
if (s->wrap > 0) s->wrap = -s->wrap; /* write the trailer only once! */
695
return s->pending != 0 ? Z_OK : Z_STREAM_END;
698
/* ========================================================================= */
699
int ZEXPORT deflateEnd (strm)
704
if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR;
706
status = strm->state->status;
707
if (status != INIT_STATE && status != BUSY_STATE &&
708
status != FINISH_STATE) {
709
return Z_STREAM_ERROR;
712
/* Deallocate in reverse order of allocations: */
713
TRY_FREE(strm, strm->state->pending_buf);
714
TRY_FREE(strm, strm->state->head);
715
TRY_FREE(strm, strm->state->prev);
716
TRY_FREE(strm, strm->state->window);
718
ZFREE(strm, strm->state);
719
strm->state = Z_NULL;
721
return status == BUSY_STATE ? Z_DATA_ERROR : Z_OK;
724
/* =========================================================================
725
* Copy the source state to the destination state.
726
* To simplify the source, this is not supported for 16-bit MSDOS (which
727
* doesn't have enough memory anyway to duplicate compression states).
729
int ZEXPORT deflateCopy (dest, source)
734
return Z_STREAM_ERROR;
741
if (source == Z_NULL || dest == Z_NULL || source->state == Z_NULL) {
742
return Z_STREAM_ERROR;
749
ds = (deflate_state *) ZALLOC(dest, 1, sizeof(deflate_state));
750
if (ds == Z_NULL) return Z_MEM_ERROR;
751
dest->state = (struct internal_state FAR *) ds;
755
ds->window = (Bytef *) ZALLOC(dest, ds->w_size, 2*sizeof(Byte));
756
ds->prev = (Posf *) ZALLOC(dest, ds->w_size, sizeof(Pos));
757
ds->head = (Posf *) ZALLOC(dest, ds->hash_size, sizeof(Pos));
758
overlay = (ushf *) ZALLOC(dest, ds->lit_bufsize, sizeof(ush)+2);
759
ds->pending_buf = (uchf *) overlay;
761
if (ds->window == Z_NULL || ds->prev == Z_NULL || ds->head == Z_NULL ||
762
ds->pending_buf == Z_NULL) {
766
/* following zmemcpy do not work for 16-bit MSDOS */
767
zmemcpy(ds->window, ss->window, ds->w_size * 2 * sizeof(Byte));
768
zmemcpy(ds->prev, ss->prev, ds->w_size * sizeof(Pos));
769
zmemcpy(ds->head, ss->head, ds->hash_size * sizeof(Pos));
770
zmemcpy(ds->pending_buf, ss->pending_buf, (uInt)ds->pending_buf_size);
772
ds->pending_out = ds->pending_buf + (ss->pending_out - ss->pending_buf);
773
ds->d_buf = overlay + ds->lit_bufsize/sizeof(ush);
774
ds->l_buf = ds->pending_buf + (1+sizeof(ush))*ds->lit_bufsize;
776
ds->l_desc.dyn_tree = ds->dyn_ltree;
777
ds->d_desc.dyn_tree = ds->dyn_dtree;
778
ds->bl_desc.dyn_tree = ds->bl_tree;
781
#endif /* MAXSEG_64K */
784
/* ===========================================================================
785
* Read a new buffer from the current input stream, update the adler32
786
* and total number of bytes read. All deflate() input goes through
787
* this function so some applications may wish to modify it to avoid
788
* allocating a large strm->next_in buffer and copying from it.
789
* (See also flush_pending()).
791
local int read_buf(strm, buf, size)
796
unsigned len = strm->avail_in;
798
if (len > size) len = size;
799
if (len == 0) return 0;
801
strm->avail_in -= len;
803
if (strm->state->wrap == 1) {
804
strm->adler = adler32(strm->adler, strm->next_in, len);
807
else if (strm->state->wrap == 2) {
808
strm->adler = crc32(strm->adler, strm->next_in, len);
811
zmemcpy(buf, strm->next_in, len);
812
strm->next_in += len;
813
strm->total_in += len;
818
/* ===========================================================================
819
* Initialize the "longest match" routines for a new zlib stream
821
local void lm_init (s)
824
s->window_size = (ulg)2L*s->w_size;
828
/* Set the default configuration parameters:
830
s->max_lazy_match = configuration_table[s->level].max_lazy;
831
s->good_match = configuration_table[s->level].good_length;
832
s->nice_match = configuration_table[s->level].nice_length;
833
s->max_chain_length = configuration_table[s->level].max_chain;
838
s->match_length = s->prev_length = MIN_MATCH-1;
839
s->match_available = 0;
842
match_init(); /* initialize the asm code */
847
/* ===========================================================================
848
* Set match_start to the longest match starting at the given string and
849
* return its length. Matches shorter or equal to prev_length are discarded,
850
* in which case the result is equal to prev_length and match_start is
852
* IN assertions: cur_match is the head of the hash chain for the current
853
* string (strstart) and its distance is <= MAX_DIST, and prev_length >= 1
854
* OUT assertion: the match length is not greater than s->lookahead.
857
/* For 80x86 and 680x0, an optimized version will be provided in match.asm or
858
* match.S. The code will be functionally equivalent.
860
local uInt longest_match(s, cur_match)
862
IPos cur_match; /* current match */
864
unsigned chain_length = s->max_chain_length;/* max hash chain length */
865
register Bytef *scan = s->window + s->strstart; /* current string */
866
register Bytef *match; /* matched string */
867
register int len; /* length of current match */
868
int best_len = s->prev_length; /* best match length so far */
869
int nice_match = s->nice_match; /* stop if match long enough */
870
IPos limit = s->strstart > (IPos)MAX_DIST(s) ?
871
s->strstart - (IPos)MAX_DIST(s) : NIL;
872
/* Stop when cur_match becomes <= limit. To simplify the code,
873
* we prevent matches with the string of window index 0.
875
Posf *prev = s->prev;
876
uInt wmask = s->w_mask;
879
/* Compare two bytes at a time. Note: this is not always beneficial.
880
* Try with and without -DUNALIGNED_OK to check.
882
register Bytef *strend = s->window + s->strstart + MAX_MATCH - 1;
883
register ush scan_start = *(ushf*)scan;
884
register ush scan_end = *(ushf*)(scan+best_len-1);
886
register Bytef *strend = s->window + s->strstart + MAX_MATCH;
887
register Byte scan_end1 = scan[best_len-1];
888
register Byte scan_end = scan[best_len];
891
/* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16.
892
* It is easy to get rid of this optimization if necessary.
894
Assert(s->hash_bits >= 8 && MAX_MATCH == 258, "Code too clever");
896
/* Do not waste too much time if we already have a good match: */
897
if (s->prev_length >= s->good_match) {
900
/* Do not look for matches beyond the end of the input. This is necessary
901
* to make deflate deterministic.
903
if ((uInt)nice_match > s->lookahead) nice_match = s->lookahead;
905
Assert((ulg)s->strstart <= s->window_size-MIN_LOOKAHEAD, "need lookahead");
908
Assert(cur_match < s->strstart, "no future");
909
match = s->window + cur_match;
911
/* Skip to next match if the match length cannot increase
912
* or if the match length is less than 2:
914
#if (defined(UNALIGNED_OK) && MAX_MATCH == 258)
915
/* This code assumes sizeof(unsigned short) == 2. Do not use
916
* UNALIGNED_OK if your compiler uses a different size.
918
if (*(ushf*)(match+best_len-1) != scan_end ||
919
*(ushf*)match != scan_start) continue;
921
/* It is not necessary to compare scan[2] and match[2] since they are
922
* always equal when the other bytes match, given that the hash keys
923
* are equal and that HASH_BITS >= 8. Compare 2 bytes at a time at
924
* strstart+3, +5, ... up to strstart+257. We check for insufficient
925
* lookahead only every 4th comparison; the 128th check will be made
926
* at strstart+257. If MAX_MATCH-2 is not a multiple of 8, it is
927
* necessary to put more guard bytes at the end of the window, or
928
* to check more often for insufficient lookahead.
930
Assert(scan[2] == match[2], "scan[2]?");
933
} while (*(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
934
*(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
935
*(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
936
*(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
938
/* The funny "do {}" generates better code on most compilers */
940
/* Here, scan <= window+strstart+257 */
941
Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan");
942
if (*scan == *match) scan++;
944
len = (MAX_MATCH - 1) - (int)(strend-scan);
945
scan = strend - (MAX_MATCH-1);
947
#else /* UNALIGNED_OK */
949
if (match[best_len] != scan_end ||
950
match[best_len-1] != scan_end1 ||
952
*++match != scan[1]) continue;
954
/* The check at best_len-1 can be removed because it will be made
955
* again later. (This heuristic is not always a win.)
956
* It is not necessary to compare scan[2] and match[2] since they
957
* are always equal when the other bytes match, given that
958
* the hash keys are equal and that HASH_BITS >= 8.
961
Assert(*scan == *match, "match[2]?");
963
/* We check for insufficient lookahead only every 8th comparison;
964
* the 256th check will be made at strstart+258.
967
} while (*++scan == *++match && *++scan == *++match &&
968
*++scan == *++match && *++scan == *++match &&
969
*++scan == *++match && *++scan == *++match &&
970
*++scan == *++match && *++scan == *++match &&
973
Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan");
975
len = MAX_MATCH - (int)(strend - scan);
976
scan = strend - MAX_MATCH;
978
#endif /* UNALIGNED_OK */
980
if (len > best_len) {
981
s->match_start = cur_match;
983
if (len >= nice_match) break;
985
scan_end = *(ushf*)(scan+best_len-1);
987
scan_end1 = scan[best_len-1];
988
scan_end = scan[best_len];
991
} while ((cur_match = prev[cur_match & wmask]) > limit
992
&& --chain_length != 0);
994
if ((uInt)best_len <= s->lookahead) return (uInt)best_len;
1000
/* ---------------------------------------------------------------------------
1001
* Optimized version for level == 1 or strategy == Z_RLE only
1003
local uInt longest_match_fast(s, cur_match)
1005
IPos cur_match; /* current match */
1007
register Bytef *scan = s->window + s->strstart; /* current string */
1008
register Bytef *match; /* matched string */
1009
register int len; /* length of current match */
1010
register Bytef *strend = s->window + s->strstart + MAX_MATCH;
1012
/* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16.
1013
* It is easy to get rid of this optimization if necessary.
1015
Assert(s->hash_bits >= 8 && MAX_MATCH == 258, "Code too clever");
1017
Assert((ulg)s->strstart <= s->window_size-MIN_LOOKAHEAD, "need lookahead");
1019
Assert(cur_match < s->strstart, "no future");
1021
match = s->window + cur_match;
1023
/* Return failure if the match length is less than 2:
1025
if (match[0] != scan[0] || match[1] != scan[1]) return MIN_MATCH-1;
1027
/* The check at best_len-1 can be removed because it will be made
1028
* again later. (This heuristic is not always a win.)
1029
* It is not necessary to compare scan[2] and match[2] since they
1030
* are always equal when the other bytes match, given that
1031
* the hash keys are equal and that HASH_BITS >= 8.
1033
scan += 2, match += 2;
1034
Assert(*scan == *match, "match[2]?");
1036
/* We check for insufficient lookahead only every 8th comparison;
1037
* the 256th check will be made at strstart+258.
1040
} while (*++scan == *++match && *++scan == *++match &&
1041
*++scan == *++match && *++scan == *++match &&
1042
*++scan == *++match && *++scan == *++match &&
1043
*++scan == *++match && *++scan == *++match &&
1046
Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan");
1048
len = MAX_MATCH - (int)(strend - scan);
1050
if (len < MIN_MATCH) return MIN_MATCH - 1;
1052
s->match_start = cur_match;
1053
return (uInt)len <= s->lookahead ? (uInt)len : s->lookahead;
1057
/* ===========================================================================
1058
* Check that the match at match_start is indeed a match.
1060
local void check_match(s, start, match, length)
1065
/* check that the match is indeed a match */
1066
if (zmemcmp(s->window + match,
1067
s->window + start, length) != EQUAL) {
1068
fprintf(stderr, " start %u, match %u, length %d\n",
1069
start, match, length);
1071
fprintf(stderr, "%c%c", s->window[match++], s->window[start++]);
1072
} while (--length != 0);
1073
z_error("invalid match");
1075
if (z_verbose > 1) {
1076
fprintf(stderr,"\\[%d,%d]", start-match, length);
1077
do { putc(s->window[start++], stderr); } while (--length != 0);
1081
# define check_match(s, start, match, length)
1084
/* ===========================================================================
1085
* Fill the window when the lookahead becomes insufficient.
1086
* Updates strstart and lookahead.
1088
* IN assertion: lookahead < MIN_LOOKAHEAD
1089
* OUT assertions: strstart <= window_size-MIN_LOOKAHEAD
1090
* At least one byte has been read, or avail_in == 0; reads are
1091
* performed for at least two bytes (required for the zip translate_eol
1092
* option -- not supported here).
1094
local void fill_window(s)
1097
register unsigned n, m;
1099
unsigned more; /* Amount of free space at the end of the window. */
1100
uInt wsize = s->w_size;
1103
more = (unsigned)(s->window_size -(ulg)s->lookahead -(ulg)s->strstart);
1105
/* Deal with !@#$% 64K limit: */
1106
if (sizeof(int) <= 2) {
1107
if (more == 0 && s->strstart == 0 && s->lookahead == 0) {
1110
} else if (more == (unsigned)(-1)) {
1111
/* Very unlikely, but possible on 16 bit machine if
1112
* strstart == 0 && lookahead == 1 (input done a byte at time)
1118
/* If the window is almost full and there is insufficient lookahead,
1119
* move the upper half to the lower one to make room in the upper half.
1121
if (s->strstart >= wsize+MAX_DIST(s)) {
1123
zmemcpy(s->window, s->window+wsize, (unsigned)wsize);
1124
s->match_start -= wsize;
1125
s->strstart -= wsize; /* we now have strstart >= MAX_DIST */
1126
s->block_start -= (long) wsize;
1128
/* Slide the hash table (could be avoided with 32 bit values
1129
at the expense of memory usage). We slide even when level == 0
1130
to keep the hash table consistent if we switch back to level > 0
1131
later. (Using level 0 permanently is not an optimal usage of
1132
zlib, so we don't care about this pathological case.)
1138
*p = (Pos)(m >= wsize ? m-wsize : NIL);
1146
*p = (Pos)(m >= wsize ? m-wsize : NIL);
1147
/* If n is not on any hash chain, prev[n] is garbage but
1148
* its value will never be used.
1154
if (s->strm->avail_in == 0) return;
1156
/* If there was no sliding:
1157
* strstart <= WSIZE+MAX_DIST-1 && lookahead <= MIN_LOOKAHEAD - 1 &&
1158
* more == window_size - lookahead - strstart
1159
* => more >= window_size - (MIN_LOOKAHEAD-1 + WSIZE + MAX_DIST-1)
1160
* => more >= window_size - 2*WSIZE + 2
1161
* In the BIG_MEM or MMAP case (not yet supported),
1162
* window_size == input_size + MIN_LOOKAHEAD &&
1163
* strstart + s->lookahead <= input_size => more >= MIN_LOOKAHEAD.
1164
* Otherwise, window_size == 2*WSIZE so more >= 2.
1165
* If there was sliding, more >= WSIZE. So in all cases, more >= 2.
1167
Assert(more >= 2, "more < 2");
1169
n = read_buf(s->strm, s->window + s->strstart + s->lookahead, more);
1172
/* Initialize the hash value now that we have some input: */
1173
if (s->lookahead >= MIN_MATCH) {
1174
s->ins_h = s->window[s->strstart];
1175
UPDATE_HASH(s, s->ins_h, s->window[s->strstart+1]);
1177
Call UPDATE_HASH() MIN_MATCH-3 more times
1180
/* If the whole input has less than MIN_MATCH bytes, ins_h is garbage,
1181
* but this is not important since only literal bytes will be emitted.
1184
} while (s->lookahead < MIN_LOOKAHEAD && s->strm->avail_in != 0);
1187
/* ===========================================================================
1188
* Flush the current block, with given end-of-file flag.
1189
* IN assertion: strstart is set to the end of the current match.
1191
#define FLUSH_BLOCK_ONLY(s, eof) { \
1192
_tr_flush_block(s, (s->block_start >= 0L ? \
1193
(charf *)&s->window[(unsigned)s->block_start] : \
1195
(ulg)((long)s->strstart - s->block_start), \
1197
s->block_start = s->strstart; \
1198
flush_pending(s->strm); \
1199
Tracev((stderr,"[FLUSH]")); \
1202
/* Same but force premature exit if necessary. */
1203
#define FLUSH_BLOCK(s, eof) { \
1204
FLUSH_BLOCK_ONLY(s, eof); \
1205
if (s->strm->avail_out == 0) return (eof) ? finish_started : need_more; \
1208
/* ===========================================================================
1209
* Copy without compression as much as possible from the input stream, return
1210
* the current block state.
1211
* This function does not insert new strings in the dictionary since
1212
* uncompressible data is probably not useful. This function is used
1213
* only for the level=0 compression option.
1214
* NOTE: this function should be optimized to avoid extra copying from
1215
* window to pending_buf.
1217
local block_state deflate_stored(s, flush)
1221
/* Stored blocks are limited to 0xffff bytes, pending_buf is limited
1222
* to pending_buf_size, and each stored block has a 5 byte header:
1224
ulg max_block_size = 0xffff;
1227
if (max_block_size > s->pending_buf_size - 5) {
1228
max_block_size = s->pending_buf_size - 5;
1231
/* Copy as much as possible from input to output: */
1233
/* Fill the window as much as possible: */
1234
if (s->lookahead <= 1) {
1236
Assert(s->strstart < s->w_size+MAX_DIST(s) ||
1237
s->block_start >= (long)s->w_size, "slide too late");
1240
if (s->lookahead == 0 && flush == Z_NO_FLUSH) return need_more;
1242
if (s->lookahead == 0) break; /* flush the current block */
1244
Assert(s->block_start >= 0L, "block gone");
1246
s->strstart += s->lookahead;
1249
/* Emit a stored block if pending_buf will be full: */
1250
max_start = s->block_start + max_block_size;
1251
if (s->strstart == 0 || (ulg)s->strstart >= max_start) {
1252
/* strstart == 0 is possible when wraparound on 16-bit machine */
1253
s->lookahead = (uInt)(s->strstart - max_start);
1254
s->strstart = (uInt)max_start;
1257
/* Flush if we may have to slide, otherwise block_start may become
1258
* negative and the data will be gone:
1260
if (s->strstart - (uInt)s->block_start >= MAX_DIST(s)) {
1264
FLUSH_BLOCK(s, flush == Z_FINISH);
1265
return flush == Z_FINISH ? finish_done : block_done;
1268
/* ===========================================================================
1269
* Compress as much as possible from the input stream, return the current
1271
* This function does not perform lazy evaluation of matches and inserts
1272
* new strings in the dictionary only for unmatched strings or for short
1273
* matches. It is used only for the fast compression options.
1275
local block_state deflate_fast(s, flush)
1279
IPos hash_head = NIL; /* head of the hash chain */
1280
int bflush; /* set if current block must be flushed */
1283
/* Make sure that we always have enough lookahead, except
1284
* at the end of the input file. We need MAX_MATCH bytes
1285
* for the next match, plus MIN_MATCH bytes to insert the
1286
* string following the next match.
1288
if (s->lookahead < MIN_LOOKAHEAD) {
1290
if (s->lookahead < MIN_LOOKAHEAD && flush == Z_NO_FLUSH) {
1293
if (s->lookahead == 0) break; /* flush the current block */
1296
/* Insert the string window[strstart .. strstart+2] in the
1297
* dictionary, and set hash_head to the head of the hash chain:
1299
if (s->lookahead >= MIN_MATCH) {
1300
INSERT_STRING(s, s->strstart, hash_head);
1303
/* Find the longest match, discarding those <= prev_length.
1304
* At this point we have always match_length < MIN_MATCH
1306
if (hash_head != NIL && s->strstart - hash_head <= MAX_DIST(s)) {
1307
/* To simplify the code, we prevent matches with the string
1308
* of window index 0 (in particular we have to avoid a match
1309
* of the string with itself at the start of the input file).
1312
if ((s->strategy < Z_HUFFMAN_ONLY) ||
1313
(s->strategy == Z_RLE && s->strstart - hash_head == 1)) {
1314
s->match_length = longest_match_fast (s, hash_head);
1317
if (s->strategy < Z_HUFFMAN_ONLY) {
1318
s->match_length = longest_match (s, hash_head);
1319
} else if (s->strategy == Z_RLE && s->strstart - hash_head == 1) {
1320
s->match_length = longest_match_fast (s, hash_head);
1323
/* longest_match() or longest_match_fast() sets match_start */
1325
if (s->match_length >= MIN_MATCH) {
1326
check_match(s, s->strstart, s->match_start, s->match_length);
1328
_tr_tally_dist(s, s->strstart - s->match_start,
1329
s->match_length - MIN_MATCH, bflush);
1331
s->lookahead -= s->match_length;
1333
/* Insert new strings in the hash table only if the match length
1334
* is not too large. This saves time but degrades compression.
1337
if (s->match_length <= s->max_insert_length &&
1338
s->lookahead >= MIN_MATCH) {
1339
s->match_length--; /* string at strstart already in table */
1342
INSERT_STRING(s, s->strstart, hash_head);
1343
/* strstart never exceeds WSIZE-MAX_MATCH, so there are
1344
* always MIN_MATCH bytes ahead.
1346
} while (--s->match_length != 0);
1351
s->strstart += s->match_length;
1352
s->match_length = 0;
1353
s->ins_h = s->window[s->strstart];
1354
UPDATE_HASH(s, s->ins_h, s->window[s->strstart+1]);
1356
Call UPDATE_HASH() MIN_MATCH-3 more times
1358
/* If lookahead < MIN_MATCH, ins_h is garbage, but it does not
1359
* matter since it will be recomputed at next deflate call.
1363
/* No match, output a literal byte */
1364
Tracevv((stderr,"%c", s->window[s->strstart]));
1365
_tr_tally_lit (s, s->window[s->strstart], bflush);
1369
if (bflush) FLUSH_BLOCK(s, 0);
1371
FLUSH_BLOCK(s, flush == Z_FINISH);
1372
return flush == Z_FINISH ? finish_done : block_done;
1376
/* ===========================================================================
1377
* Same as above, but achieves better compression. We use a lazy
1378
* evaluation for matches: a match is finally adopted only if there is
1379
* no better match at the next window position.
1381
local block_state deflate_slow(s, flush)
1385
IPos hash_head = NIL; /* head of hash chain */
1386
int bflush; /* set if current block must be flushed */
1388
/* Process the input block. */
1390
/* Make sure that we always have enough lookahead, except
1391
* at the end of the input file. We need MAX_MATCH bytes
1392
* for the next match, plus MIN_MATCH bytes to insert the
1393
* string following the next match.
1395
if (s->lookahead < MIN_LOOKAHEAD) {
1397
if (s->lookahead < MIN_LOOKAHEAD && flush == Z_NO_FLUSH) {
1400
if (s->lookahead == 0) break; /* flush the current block */
1403
/* Insert the string window[strstart .. strstart+2] in the
1404
* dictionary, and set hash_head to the head of the hash chain:
1406
if (s->lookahead >= MIN_MATCH) {
1407
INSERT_STRING(s, s->strstart, hash_head);
1410
/* Find the longest match, discarding those <= prev_length.
1412
s->prev_length = s->match_length, s->prev_match = s->match_start;
1413
s->match_length = MIN_MATCH-1;
1415
if (hash_head != NIL && s->prev_length < s->max_lazy_match &&
1416
s->strstart - hash_head <= MAX_DIST(s)) {
1417
/* To simplify the code, we prevent matches with the string
1418
* of window index 0 (in particular we have to avoid a match
1419
* of the string with itself at the start of the input file).
1421
if (s->strategy < Z_HUFFMAN_ONLY) {
1422
s->match_length = longest_match (s, hash_head);
1423
} else if (s->strategy == Z_RLE && s->strstart - hash_head == 1) {
1424
s->match_length = longest_match_fast (s, hash_head);
1426
/* longest_match() or longest_match_fast() sets match_start */
1428
if (s->match_length <= 5 && (s->strategy == Z_FILTERED
1429
#if TOO_FAR <= 32767
1430
|| (s->match_length == MIN_MATCH &&
1431
s->strstart - s->match_start > TOO_FAR)
1435
/* If prev_match is also MIN_MATCH, match_start is garbage
1436
* but we will ignore the current match anyway.
1438
s->match_length = MIN_MATCH-1;
1441
/* If there was a match at the previous step and the current
1442
* match is not better, output the previous match:
1444
if (s->prev_length >= MIN_MATCH && s->match_length <= s->prev_length) {
1445
uInt max_insert = s->strstart + s->lookahead - MIN_MATCH;
1446
/* Do not insert strings in hash table beyond this. */
1448
check_match(s, s->strstart-1, s->prev_match, s->prev_length);
1450
_tr_tally_dist(s, s->strstart -1 - s->prev_match,
1451
s->prev_length - MIN_MATCH, bflush);
1453
/* Insert in hash table all strings up to the end of the match.
1454
* strstart-1 and strstart are already inserted. If there is not
1455
* enough lookahead, the last two strings are not inserted in
1458
s->lookahead -= s->prev_length-1;
1459
s->prev_length -= 2;
1461
if (++s->strstart <= max_insert) {
1462
INSERT_STRING(s, s->strstart, hash_head);
1464
} while (--s->prev_length != 0);
1465
s->match_available = 0;
1466
s->match_length = MIN_MATCH-1;
1469
if (bflush) FLUSH_BLOCK(s, 0);
1471
} else if (s->match_available) {
1472
/* If there was no match at the previous position, output a
1473
* single literal. If there was a match but the current match
1474
* is longer, truncate the previous match to a single literal.
1476
Tracevv((stderr,"%c", s->window[s->strstart-1]));
1477
_tr_tally_lit(s, s->window[s->strstart-1], bflush);
1479
FLUSH_BLOCK_ONLY(s, 0);
1483
if (s->strm->avail_out == 0) return need_more;
1485
/* There is no previous match to compare with, wait for
1486
* the next step to decide.
1488
s->match_available = 1;
1493
Assert (flush != Z_NO_FLUSH, "no flush?");
1494
if (s->match_available) {
1495
Tracevv((stderr,"%c", s->window[s->strstart-1]));
1496
_tr_tally_lit(s, s->window[s->strstart-1], bflush);
1497
s->match_available = 0;
1499
FLUSH_BLOCK(s, flush == Z_FINISH);
1500
return flush == Z_FINISH ? finish_done : block_done;
1502
#endif /* FASTEST */
added
removed
utils/TkCximage/src/zlib/deflate.c
