1 /*
2 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
3 *
4 * This code is free software; you can redistribute it and/or modify it
5 * under the terms of the GNU General Public License version 2 only, as
6 * published by the Free Software Foundation. Oracle designates this
7 * particular file as subject to the "Classpath" exception as provided
8 * by Oracle in the LICENSE file that accompanied this code.
9 *
10 * This code is distributed in the hope that it will be useful, but WITHOUT
11 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
12 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
13 * version 2 for more details (a copy is included in the LICENSE file that
14 * accompanied this code).
15 *
16 * You should have received a copy of the GNU General Public License version
17 * 2 along with this work; if not, write to the Free Software Foundation,
18 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
19 *
20 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
21 * or visit www.oracle.com if you need additional information or have any
22 * questions.
23 */
24
25 /* deflate.c -- compress data using the deflation algorithm
26 * Copyright (C) 1995-2013 Jean-loup Gailly and Mark Adler
27 * For conditions of distribution and use, see copyright notice in zlib.h
28 */
29
30 /*
31 * ALGORITHM
32 *
33 * The "deflation" process depends on being able to identify portions
34 * of the input text which are identical to earlier input (within a
35 * sliding window trailing behind the input currently being processed).
36 *
37 * The most straightforward technique turns out to be the fastest for
38 * most input files: try all possible matches and select the longest.
39 * The key feature of this algorithm is that insertions into the string
40 * dictionary are very simple and thus fast, and deletions are avoided
41 * completely. Insertions are performed at each input character, whereas
42 * string matches are performed only when the previous match ends. So it
43 * is preferable to spend more time in matches to allow very fast string
44 * insertions and avoid deletions. The matching algorithm for small
45 * strings is inspired from that of Rabin & Karp. A brute force approach
46 * is used to find longer strings when a small match has been found.
47 * A similar algorithm is used in comic (by Jan-Mark Wams) and freeze
48 * (by Leonid Broukhis).
49 * A previous version of this file used a more sophisticated algorithm
50 * (by Fiala and Greene) which is guaranteed to run in linear amortized
51 * time, but has a larger average cost, uses more memory and is patented.
52 * However the F&G algorithm may be faster for some highly redundant
53 * files if the parameter max_chain_length (described below) is too large.
54 *
55 * ACKNOWLEDGEMENTS
56 *
57 * The idea of lazy evaluation of matches is due to Jan-Mark Wams, and
58 * I found it in 'freeze' written by Leonid Broukhis.
59 * Thanks to many people for bug reports and testing.
60 *
61 * REFERENCES
62 *
63 * Deutsch, L.P.,"DEFLATE Compressed Data Format Specification".
64 * Available in http://tools.ietf.org/html/rfc1951
65 *
66 * A description of the Rabin and Karp algorithm is given in the book
67 * "Algorithms" by R. Sedgewick, Addison-Wesley, p252.
68 *
69 * Fiala,E.R., and Greene,D.H.
70 * Data Compression with Finite Windows, Comm.ACM, 32,4 (1989) 490-595
71 *
72 */
73
74 /* @(#) $Id$ */
75
76 #include "deflate.h"
77
78 const char deflate_copyright[] =
79 " deflate 1.2.8 Copyright 1995-2013 Jean-loup Gailly and Mark Adler ";
80 /*
81 If you use the zlib library in a product, an acknowledgment is welcome
82 in the documentation of your product. If for some reason you cannot
83 include such an acknowledgment, I would appreciate that you keep this
84 copyright string in the executable of your product.
85 */
86
87 /* ===========================================================================
88 * Function prototypes.
89 */
90 typedef enum {
91 need_more, /* block not completed, need more input or more output */
92 block_done, /* block flush performed */
93 finish_started, /* finish started, need only more output at next deflate */
94 finish_done /* finish done, accept no more input or output */
95 } block_state;
96
97 typedef block_state (*compress_func) OF((deflate_state *s, int flush));
98 /* Compression function. Returns the block state after the call. */
99
100 local void fill_window OF((deflate_state *s));
101 local block_state deflate_stored OF((deflate_state *s, int flush));
102 local block_state deflate_fast OF((deflate_state *s, int flush));
103 #ifndef FASTEST
104 local block_state deflate_slow OF((deflate_state *s, int flush));
105 #endif
106 local block_state deflate_rle OF((deflate_state *s, int flush));
107 local block_state deflate_huff OF((deflate_state *s, int flush));
108 local void lm_init OF((deflate_state *s));
109 local void putShortMSB OF((deflate_state *s, uInt b));
110 local void flush_pending OF((z_streamp strm));
111 local int read_buf OF((z_streamp strm, Bytef *buf, unsigned size));
112 #ifdef ASMV
113 void match_init OF((void)); /* asm code initialization */
114 uInt longest_match OF((deflate_state *s, IPos cur_match));
115 #else
116 local uInt longest_match OF((deflate_state *s, IPos cur_match));
117 #endif
118
119 #ifdef DEBUG
120 local void check_match OF((deflate_state *s, IPos start, IPos match,
121 int length));
122 #endif
123
124 /* ===========================================================================
125 * Local data
126 */
127
128 #define NIL 0
129 /* Tail of hash chains */
130
131 #ifndef TOO_FAR
132 # define TOO_FAR 4096
133 #endif
134 /* Matches of length 3 are discarded if their distance exceeds TOO_FAR */
135
136 /* Values for max_lazy_match, good_match and max_chain_length, depending on
137 * the desired pack level (0..9). The values given below have been tuned to
138 * exclude worst case performance for pathological files. Better values may be
139 * found for specific files.
140 */
141 typedef struct config_s {
142 ush good_length; /* reduce lazy search above this match length */
143 ush max_lazy; /* do not perform lazy search above this match length */
144 ush nice_length; /* quit search above this match length */
145 ush max_chain;
146 compress_func func;
147 } config;
148
149 #ifdef FASTEST
150 local const config configuration_table[2] = {
151 /* good lazy nice chain */
152 /* 0 */ {0, 0, 0, 0, deflate_stored}, /* store only */
153 /* 1 */ {4, 4, 8, 4, deflate_fast}}; /* max speed, no lazy matches */
154 #else
155 local const config configuration_table[10] = {
156 /* good lazy nice chain */
157 /* 0 */ {0, 0, 0, 0, deflate_stored}, /* store only */
158 /* 1 */ {4, 4, 8, 4, deflate_fast}, /* max speed, no lazy matches */
159 /* 2 */ {4, 5, 16, 8, deflate_fast},
160 /* 3 */ {4, 6, 32, 32, deflate_fast},
161
162 /* 4 */ {4, 4, 16, 16, deflate_slow}, /* lazy matches */
163 /* 5 */ {8, 16, 32, 32, deflate_slow},
164 /* 6 */ {8, 16, 128, 128, deflate_slow},
165 /* 7 */ {8, 32, 128, 256, deflate_slow},
166 /* 8 */ {32, 128, 258, 1024, deflate_slow},
167 /* 9 */ {32, 258, 258, 4096, deflate_slow}}; /* max compression */
168 #endif
169
170 /* Note: the deflate() code requires max_lazy >= MIN_MATCH and max_chain >= 4
171 * For deflate_fast() (levels <= 3) good is ignored and lazy has a different
172 * meaning.
173 */
174
175 #define EQUAL 0
176 /* result of memcmp for equal strings */
177
178 #ifndef NO_DUMMY_DECL
179 struct static_tree_desc_s {int dummy;}; /* for buggy compilers */
180 #endif
181
182 /* rank Z_BLOCK between Z_NO_FLUSH and Z_PARTIAL_FLUSH */
183 #define RANK(f) (((f) << 1) - ((f) > 4 ? 9 : 0))
184
185 /* ===========================================================================
186 * Update a hash value with the given input byte
187 * IN assertion: all calls to to UPDATE_HASH are made with consecutive
188 * input characters, so that a running hash key can be computed from the
189 * previous key instead of complete recalculation each time.
190 */
191 #define UPDATE_HASH(s,h,c) (h = (((h)<<s->hash_shift) ^ (c)) & s->hash_mask)
192
193
194 /* ===========================================================================
195 * Insert string str in the dictionary and set match_head to the previous head
196 * of the hash chain (the most recent string with same hash key). Return
197 * the previous length of the hash chain.
198 * If this file is compiled with -DFASTEST, the compression level is forced
199 * to 1, and no hash chains are maintained.
200 * IN assertion: all calls to to INSERT_STRING are made with consecutive
201 * input characters and the first MIN_MATCH bytes of str are valid
202 * (except for the last MIN_MATCH-1 bytes of the input file).
203 */
204 #ifdef FASTEST
205 #define INSERT_STRING(s, str, match_head) \
206 (UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \
207 match_head = s->head[s->ins_h], \
208 s->head[s->ins_h] = (Pos)(str))
209 #else
210 #define INSERT_STRING(s, str, match_head) \
211 (UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \
212 match_head = s->prev[(str) & s->w_mask] = s->head[s->ins_h], \
213 s->head[s->ins_h] = (Pos)(str))
214 #endif
215
216 /* ===========================================================================
217 * Initialize the hash table (avoiding 64K overflow for 16 bit systems).
218 * prev[] will be initialized on the fly.
219 */
220 #define CLEAR_HASH(s) \
221 s->head[s->hash_size-1] = NIL; \
222 zmemzero((Bytef *)s->head, (unsigned)(s->hash_size-1)*sizeof(*s->head));
223
224 /* ========================================================================= */
225 int ZEXPORT deflateInit_(strm, level, version, stream_size)
226 z_streamp strm;
227 int level;
228 const char *version;
229 int stream_size;
230 {
231 return deflateInit2_(strm, level, Z_DEFLATED, MAX_WBITS, DEF_MEM_LEVEL,
232 Z_DEFAULT_STRATEGY, version, stream_size);
233 /* To do: ignore strm->next_in if we use it as window */
234 }
235
236 /* ========================================================================= */
237 int ZEXPORT deflateInit2_(strm, level, method, windowBits, memLevel, strategy,
238 version, stream_size)
239 z_streamp strm;
240 int level;
241 int method;
242 int windowBits;
243 int memLevel;
244 int strategy;
245 const char *version;
246 int stream_size;
247 {
248 deflate_state *s;
249 int wrap = 1;
250 static const char my_version[] = ZLIB_VERSION;
251
252 ushf *overlay;
253 /* We overlay pending_buf and d_buf+l_buf. This works since the average
254 * output size for (length,distance) codes is <= 24 bits.
255 */
256
257 if (version == Z_NULL || version[0] != my_version[0] ||
258 stream_size != sizeof(z_stream)) {
259 return Z_VERSION_ERROR;
260 }
261 if (strm == Z_NULL) return Z_STREAM_ERROR;
262
263 strm->msg = Z_NULL;
264 if (strm->zalloc == (alloc_func)0) {
265 #ifdef Z_SOLO
266 return Z_STREAM_ERROR;
267 #else
268 strm->zalloc = zcalloc;
269 strm->opaque = (voidpf)0;
270 #endif
271 }
272 if (strm->zfree == (free_func)0)
273 #ifdef Z_SOLO
274 return Z_STREAM_ERROR;
275 #else
276 strm->zfree = zcfree;
277 #endif
278
279 #ifdef FASTEST
280 if (level != 0) level = 1;
281 #else
282 if (level == Z_DEFAULT_COMPRESSION) level = 6;
283 #endif
284
285 if (windowBits < 0) { /* suppress zlib wrapper */
286 wrap = 0;
287 windowBits = -windowBits;
288 }
289 #ifdef GZIP
290 else if (windowBits > 15) {
291 wrap = 2; /* write gzip wrapper instead */
292 windowBits -= 16;
293 }
294 #endif
295 if (memLevel < 1 || memLevel > MAX_MEM_LEVEL || method != Z_DEFLATED ||
296 windowBits < 8 || windowBits > 15 || level < 0 || level > 9 ||
297 strategy < 0 || strategy > Z_FIXED) {
298 return Z_STREAM_ERROR;
299 }
300 if (windowBits == 8) windowBits = 9; /* until 256-byte window bug fixed */
301 s = (deflate_state *) ZALLOC(strm, 1, sizeof(deflate_state));
302 if (s == Z_NULL) return Z_MEM_ERROR;
303 strm->state = (struct internal_state FAR *)s;
304 s->strm = strm;
305
306 s->wrap = wrap;
307 s->gzhead = Z_NULL;
308 s->w_bits = windowBits;
309 s->w_size = 1 << s->w_bits;
310 s->w_mask = s->w_size - 1;
311
312 s->hash_bits = memLevel + 7;
313 s->hash_size = 1 << s->hash_bits;
314 s->hash_mask = s->hash_size - 1;
315 s->hash_shift = ((s->hash_bits+MIN_MATCH-1)/MIN_MATCH);
316
317 s->window = (Bytef *) ZALLOC(strm, s->w_size, 2*sizeof(Byte));
318 s->prev = (Posf *) ZALLOC(strm, s->w_size, sizeof(Pos));
319 s->head = (Posf *) ZALLOC(strm, s->hash_size, sizeof(Pos));
320
321 s->high_water = 0; /* nothing written to s->window yet */
322
323 s->lit_bufsize = 1 << (memLevel + 6); /* 16K elements by default */
324
325 overlay = (ushf *) ZALLOC(strm, s->lit_bufsize, sizeof(ush)+2);
326 s->pending_buf = (uchf *) overlay;
327 s->pending_buf_size = (ulg)s->lit_bufsize * (sizeof(ush)+2L);
328
329 if (s->window == Z_NULL || s->prev == Z_NULL || s->head == Z_NULL ||
330 s->pending_buf == Z_NULL) {
331 s->status = FINISH_STATE;
332 strm->msg = ERR_MSG(Z_MEM_ERROR);
333 deflateEnd (strm);
334 return Z_MEM_ERROR;
335 }
336 s->d_buf = overlay + s->lit_bufsize/sizeof(ush);
337 s->l_buf = s->pending_buf + (1+sizeof(ush))*s->lit_bufsize;
338
339 s->level = level;
340 s->strategy = strategy;
341 s->method = (Byte)method;
342
343 return deflateReset(strm);
344 }
345
346 /* ========================================================================= */
347 int ZEXPORT deflateSetDictionary (strm, dictionary, dictLength)
348 z_streamp strm;
349 const Bytef *dictionary;
350 uInt dictLength;
351 {
352 deflate_state *s;
353 uInt str, n;
354 int wrap;
355 unsigned avail;
356 z_const unsigned char *next;
357
358 if (strm == Z_NULL || strm->state == Z_NULL || dictionary == Z_NULL)
359 return Z_STREAM_ERROR;
360 s = strm->state;
361 wrap = s->wrap;
362 if (wrap == 2 || (wrap == 1 && s->status != INIT_STATE) || s->lookahead)
363 return Z_STREAM_ERROR;
364
365 /* when using zlib wrappers, compute Adler-32 for provided dictionary */
366 if (wrap == 1)
367 strm->adler = adler32(strm->adler, dictionary, dictLength);
368 s->wrap = 0; /* avoid computing Adler-32 in read_buf */
369
370 /* if dictionary would fill window, just replace the history */
371 if (dictLength >= s->w_size) {
372 if (wrap == 0) { /* already empty otherwise */
373 CLEAR_HASH(s);
374 s->strstart = 0;
375 s->block_start = 0L;
376 s->insert = 0;
377 }
378 dictionary += dictLength - s->w_size; /* use the tail */
379 dictLength = s->w_size;
380 }
381
382 /* insert dictionary into window and hash */
383 avail = strm->avail_in;
384 next = strm->next_in;
385 strm->avail_in = dictLength;
386 strm->next_in = (z_const Bytef *)dictionary;
387 fill_window(s);
388 while (s->lookahead >= MIN_MATCH) {
389 str = s->strstart;
390 n = s->lookahead - (MIN_MATCH-1);
391 do {
392 UPDATE_HASH(s, s->ins_h, s->window[str + MIN_MATCH-1]);
393 #ifndef FASTEST
394 s->prev[str & s->w_mask] = s->head[s->ins_h];
395 #endif
396 s->head[s->ins_h] = (Pos)str;
397 str++;
398 } while (--n);
399 s->strstart = str;
400 s->lookahead = MIN_MATCH-1;
401 fill_window(s);
402 }
403 s->strstart += s->lookahead;
404 s->block_start = (long)s->strstart;
405 s->insert = s->lookahead;
406 s->lookahead = 0;
407 s->match_length = s->prev_length = MIN_MATCH-1;
408 s->match_available = 0;
409 strm->next_in = next;
410 strm->avail_in = avail;
411 s->wrap = wrap;
412 return Z_OK;
413 }
414
415 /* ========================================================================= */
416 int ZEXPORT deflateResetKeep (strm)
417 z_streamp strm;
418 {
419 deflate_state *s;
420
421 if (strm == Z_NULL || strm->state == Z_NULL ||
422 strm->zalloc == (alloc_func)0 || strm->zfree == (free_func)0) {
423 return Z_STREAM_ERROR;
424 }
425
426 strm->total_in = strm->total_out = 0;
427 strm->msg = Z_NULL; /* use zfree if we ever allocate msg dynamically */
428 strm->data_type = Z_UNKNOWN;
429
430 s = (deflate_state *)strm->state;
431 s->pending = 0;
432 s->pending_out = s->pending_buf;
433
434 if (s->wrap < 0) {
435 s->wrap = -s->wrap; /* was made negative by deflate(..., Z_FINISH); */
436 }
437 s->status = s->wrap ? INIT_STATE : BUSY_STATE;
438 strm->adler =
439 #ifdef GZIP
440 s->wrap == 2 ? crc32(0L, Z_NULL, 0) :
441 #endif
442 adler32(0L, Z_NULL, 0);
443 s->last_flush = Z_NO_FLUSH;
444
445 _tr_init(s);
446
447 return Z_OK;
448 }
449
450 /* ========================================================================= */
451 int ZEXPORT deflateReset (strm)
452 z_streamp strm;
453 {
454 int ret;
455
456 ret = deflateResetKeep(strm);
457 if (ret == Z_OK)
458 lm_init(strm->state);
459 return ret;
460 }
461
462 /* ========================================================================= */
463 int ZEXPORT deflateSetHeader (strm, head)
464 z_streamp strm;
465 gz_headerp head;
466 {
467 if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR;
468 if (strm->state->wrap != 2) return Z_STREAM_ERROR;
469 strm->state->gzhead = head;
470 return Z_OK;
471 }
472
473 /* ========================================================================= */
474 int ZEXPORT deflatePending (strm, pending, bits)
475 unsigned *pending;
476 int *bits;
477 z_streamp strm;
478 {
479 if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR;
480 if (pending != Z_NULL)
481 *pending = strm->state->pending;
482 if (bits != Z_NULL)
483 *bits = strm->state->bi_valid;
484 return Z_OK;
485 }
486
487 /* ========================================================================= */
488 int ZEXPORT deflatePrime (strm, bits, value)
489 z_streamp strm;
490 int bits;
491 int value;
492 {
493 deflate_state *s;
494 int put;
495
496 if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR;
497 s = strm->state;
498 if ((Bytef *)(s->d_buf) < s->pending_out + ((Buf_size + 7) >> 3))
499 return Z_BUF_ERROR;
500 do {
501 put = Buf_size - s->bi_valid;
502 if (put > bits)
503 put = bits;
504 s->bi_buf |= (ush)((value & ((1 << put) - 1)) << s->bi_valid);
505 s->bi_valid += put;
506 _tr_flush_bits(s);
507 value >>= put;
508 bits -= put;
509 } while (bits);
510 return Z_OK;
511 }
512
513 /* ========================================================================= */
514 int ZEXPORT deflateParams(strm, level, strategy)
515 z_streamp strm;
516 int level;
517 int strategy;
518 {
519 deflate_state *s;
520 compress_func func;
521 int err = Z_OK;
522
523 if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR;
524 s = strm->state;
525
526 #ifdef FASTEST
527 if (level != 0) level = 1;
528 #else
529 if (level == Z_DEFAULT_COMPRESSION) level = 6;
530 #endif
531 if (level < 0 || level > 9 || strategy < 0 || strategy > Z_FIXED) {
532 return Z_STREAM_ERROR;
533 }
534 func = configuration_table[s->level].func;
535
536 if ((strategy != s->strategy || func != configuration_table[level].func) &&
537 strm->total_in != 0) {
538 /* Flush the last buffer: */
539 err = deflate(strm, Z_BLOCK);
540 if (err == Z_BUF_ERROR && s->pending == 0)
541 err = Z_OK;
542 }
543 if (s->level != level) {
544 s->level = level;
545 s->max_lazy_match = configuration_table[level].max_lazy;
546 s->good_match = configuration_table[level].good_length;
547 s->nice_match = configuration_table[level].nice_length;
548 s->max_chain_length = configuration_table[level].max_chain;
549 }
550 s->strategy = strategy;
551 return err;
552 }
553
554 /* ========================================================================= */
555 int ZEXPORT deflateTune(strm, good_length, max_lazy, nice_length, max_chain)
556 z_streamp strm;
557 int good_length;
558 int max_lazy;
559 int nice_length;
560 int max_chain;
561 {
562 deflate_state *s;
563
564 if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR;
565 s = strm->state;
566 s->good_match = good_length;
567 s->max_lazy_match = max_lazy;
568 s->nice_match = nice_length;
569 s->max_chain_length = max_chain;
570 return Z_OK;
571 }
572
573 /* =========================================================================
574 * For the default windowBits of 15 and memLevel of 8, this function returns
575 * a close to exact, as well as small, upper bound on the compressed size.
576 * They are coded as constants here for a reason--if the #define's are
577 * changed, then this function needs to be changed as well. The return
578 * value for 15 and 8 only works for those exact settings.
579 *
580 * For any setting other than those defaults for windowBits and memLevel,
581 * the value returned is a conservative worst case for the maximum expansion
582 * resulting from using fixed blocks instead of stored blocks, which deflate
583 * can emit on compressed data for some combinations of the parameters.
584 *
585 * This function could be more sophisticated to provide closer upper bounds for
586 * every combination of windowBits and memLevel. But even the conservative
587 * upper bound of about 14% expansion does not seem onerous for output buffer
588 * allocation.
589 */
590 uLong ZEXPORT deflateBound(strm, sourceLen)
591 z_streamp strm;
592 uLong sourceLen;
593 {
594 deflate_state *s;
595 uLong complen, wraplen;
596 Bytef *str;
597
598 /* conservative upper bound for compressed data */
599 complen = sourceLen +
600 ((sourceLen + 7) >> 3) + ((sourceLen + 63) >> 6) + 5;
601
602 /* if can't get parameters, return conservative bound plus zlib wrapper */
603 if (strm == Z_NULL || strm->state == Z_NULL)
604 return complen + 6;
605
606 /* compute wrapper length */
607 s = strm->state;
608 switch (s->wrap) {
609 case 0: /* raw deflate */
610 wraplen = 0;
611 break;
612 case 1: /* zlib wrapper */
613 wraplen = 6 + (s->strstart ? 4 : 0);
614 break;
615 case 2: /* gzip wrapper */
616 wraplen = 18;
617 if (s->gzhead != Z_NULL) { /* user-supplied gzip header */
618 if (s->gzhead->extra != Z_NULL)
619 wraplen += 2 + s->gzhead->extra_len;
620 str = s->gzhead->name;
621 if (str != Z_NULL)
622 do {
623 wraplen++;
624 } while (*str++);
625 str = s->gzhead->comment;
626 if (str != Z_NULL)
627 do {
628 wraplen++;
629 } while (*str++);
630 if (s->gzhead->hcrc)
631 wraplen += 2;
632 }
633 break;
634 default: /* for compiler happiness */
635 wraplen = 6;
636 }
637
638 /* if not default parameters, return conservative bound */
639 if (s->w_bits != 15 || s->hash_bits != 8 + 7)
640 return complen + wraplen;
641
642 /* default settings: return tight bound for that case */
643 return sourceLen + (sourceLen >> 12) + (sourceLen >> 14) +
644 (sourceLen >> 25) + 13 - 6 + wraplen;
645 }
646
647 /* =========================================================================
648 * Put a short in the pending buffer. The 16-bit value is put in MSB order.
649 * IN assertion: the stream state is correct and there is enough room in
650 * pending_buf.
651 */
652 local void putShortMSB (s, b)
653 deflate_state *s;
654 uInt b;
655 {
656 put_byte(s, (Byte)(b >> 8));
657 put_byte(s, (Byte)(b & 0xff));
658 }
659
660 /* =========================================================================
661 * Flush as much pending output as possible. All deflate() output goes
662 * through this function so some applications may wish to modify it
663 * to avoid allocating a large strm->next_out buffer and copying into it.
664 * (See also read_buf()).
665 */
666 local void flush_pending(strm)
667 z_streamp strm;
668 {
669 unsigned len;
670 deflate_state *s = strm->state;
671
672 _tr_flush_bits(s);
673 len = s->pending;
674 if (len > strm->avail_out) len = strm->avail_out;
675 if (len == 0) return;
676
677 zmemcpy(strm->next_out, s->pending_out, len);
678 strm->next_out += len;
679 s->pending_out += len;
680 strm->total_out += len;
681 strm->avail_out -= len;
682 s->pending -= len;
683 if (s->pending == 0) {
684 s->pending_out = s->pending_buf;
685 }
686 }
687
688 /* ========================================================================= */
689 int ZEXPORT deflate (strm, flush)
690 z_streamp strm;
691 int flush;
692 {
693 int old_flush; /* value of flush param for previous deflate call */
694 deflate_state *s;
695
696 if (strm == Z_NULL || strm->state == Z_NULL ||
697 flush > Z_BLOCK || flush < 0) {
698 return Z_STREAM_ERROR;
699 }
700 s = strm->state;
701
702 if (strm->next_out == Z_NULL ||
703 (strm->next_in == Z_NULL && strm->avail_in != 0) ||
704 (s->status == FINISH_STATE && flush != Z_FINISH)) {
705 ERR_RETURN(strm, Z_STREAM_ERROR);
706 }
707 if (strm->avail_out == 0) ERR_RETURN(strm, Z_BUF_ERROR);
708
709 s->strm = strm; /* just in case */
710 old_flush = s->last_flush;
711 s->last_flush = flush;
712
713 /* Write the header */
714 if (s->status == INIT_STATE) {
715 #ifdef GZIP
716 if (s->wrap == 2) {
717 strm->adler = crc32(0L, Z_NULL, 0);
718 put_byte(s, 31);
719 put_byte(s, 139);
720 put_byte(s, 8);
721 if (s->gzhead == Z_NULL) {
722 put_byte(s, 0);
723 put_byte(s, 0);
724 put_byte(s, 0);
725 put_byte(s, 0);
726 put_byte(s, 0);
727 put_byte(s, s->level == 9 ? 2 :
728 (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ?
729 4 : 0));
730 put_byte(s, OS_CODE);
731 s->status = BUSY_STATE;
732 }
733 else {
734 put_byte(s, (s->gzhead->text ? 1 : 0) +
735 (s->gzhead->hcrc ? 2 : 0) +
736 (s->gzhead->extra == Z_NULL ? 0 : 4) +
737 (s->gzhead->name == Z_NULL ? 0 : 8) +
738 (s->gzhead->comment == Z_NULL ? 0 : 16)
739 );
740 put_byte(s, (Byte)(s->gzhead->time & 0xff));
741 put_byte(s, (Byte)((s->gzhead->time >> 8) & 0xff));
742 put_byte(s, (Byte)((s->gzhead->time >> 16) & 0xff));
743 put_byte(s, (Byte)((s->gzhead->time >> 24) & 0xff));
744 put_byte(s, s->level == 9 ? 2 :
745 (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ?
746 4 : 0));
747 put_byte(s, s->gzhead->os & 0xff);
748 if (s->gzhead->extra != Z_NULL) {
749 put_byte(s, s->gzhead->extra_len & 0xff);
750 put_byte(s, (s->gzhead->extra_len >> 8) & 0xff);
751 }
752 if (s->gzhead->hcrc)
753 strm->adler = crc32(strm->adler, s->pending_buf,
754 s->pending);
755 s->gzindex = 0;
756 s->status = EXTRA_STATE;
757 }
758 }
759 else
760 #endif
761 {
762 uInt header = (Z_DEFLATED + ((s->w_bits-8)<<4)) << 8;
763 uInt level_flags;
764
765 if (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2)
766 level_flags = 0;
767 else if (s->level < 6)
768 level_flags = 1;
769 else if (s->level == 6)
770 level_flags = 2;
771 else
772 level_flags = 3;
773 header |= (level_flags << 6);
774 if (s->strstart != 0) header |= PRESET_DICT;
775 header += 31 - (header % 31);
776
777 s->status = BUSY_STATE;
778 putShortMSB(s, header);
779
780 /* Save the adler32 of the preset dictionary: */
781 if (s->strstart != 0) {
782 putShortMSB(s, (uInt)(strm->adler >> 16));
783 putShortMSB(s, (uInt)(strm->adler & 0xffff));
784 }
785 strm->adler = adler32(0L, Z_NULL, 0);
786 }
787 }
788 #ifdef GZIP
789 if (s->status == EXTRA_STATE) {
790 if (s->gzhead->extra != Z_NULL) {
791 uInt beg = s->pending; /* start of bytes to update crc */
792
793 while (s->gzindex < (s->gzhead->extra_len & 0xffff)) {
794 if (s->pending == s->pending_buf_size) {
795 if (s->gzhead->hcrc && s->pending > beg)
796 strm->adler = crc32(strm->adler, s->pending_buf + beg,
797 s->pending - beg);
798 flush_pending(strm);
799 beg = s->pending;
800 if (s->pending == s->pending_buf_size)
801 break;
802 }
803 put_byte(s, s->gzhead->extra[s->gzindex]);
804 s->gzindex++;
805 }
806 if (s->gzhead->hcrc && s->pending > beg)
807 strm->adler = crc32(strm->adler, s->pending_buf + beg,
808 s->pending - beg);
809 if (s->gzindex == s->gzhead->extra_len) {
810 s->gzindex = 0;
811 s->status = NAME_STATE;
812 }
813 }
814 else
815 s->status = NAME_STATE;
816 }
817 if (s->status == NAME_STATE) {
818 if (s->gzhead->name != Z_NULL) {
819 uInt beg = s->pending; /* start of bytes to update crc */
820 int val;
821
822 do {
823 if (s->pending == s->pending_buf_size) {
824 if (s->gzhead->hcrc && s->pending > beg)
825 strm->adler = crc32(strm->adler, s->pending_buf + beg,
826 s->pending - beg);
827 flush_pending(strm);
828 beg = s->pending;
829 if (s->pending == s->pending_buf_size) {
830 val = 1;
831 break;
832 }
833 }
834 val = s->gzhead->name[s->gzindex++];
835 put_byte(s, val);
836 } while (val != 0);
837 if (s->gzhead->hcrc && s->pending > beg)
838 strm->adler = crc32(strm->adler, s->pending_buf + beg,
839 s->pending - beg);
840 if (val == 0) {
841 s->gzindex = 0;
842 s->status = COMMENT_STATE;
843 }
844 }
845 else
846 s->status = COMMENT_STATE;
847 }
848 if (s->status == COMMENT_STATE) {
849 if (s->gzhead->comment != Z_NULL) {
850 uInt beg = s->pending; /* start of bytes to update crc */
851 int val;
852
853 do {
854 if (s->pending == s->pending_buf_size) {
855 if (s->gzhead->hcrc && s->pending > beg)
856 strm->adler = crc32(strm->adler, s->pending_buf + beg,
857 s->pending - beg);
858 flush_pending(strm);
859 beg = s->pending;
860 if (s->pending == s->pending_buf_size) {
861 val = 1;
862 break;
863 }
864 }
865 val = s->gzhead->comment[s->gzindex++];
866 put_byte(s, val);
867 } while (val != 0);
868 if (s->gzhead->hcrc && s->pending > beg)
869 strm->adler = crc32(strm->adler, s->pending_buf + beg,
870 s->pending - beg);
871 if (val == 0)
872 s->status = HCRC_STATE;
873 }
874 else
875 s->status = HCRC_STATE;
876 }
877 if (s->status == HCRC_STATE) {
878 if (s->gzhead->hcrc) {
879 if (s->pending + 2 > s->pending_buf_size)
880 flush_pending(strm);
881 if (s->pending + 2 <= s->pending_buf_size) {
882 put_byte(s, (Byte)(strm->adler & 0xff));
883 put_byte(s, (Byte)((strm->adler >> 8) & 0xff));
884 strm->adler = crc32(0L, Z_NULL, 0);
885 s->status = BUSY_STATE;
886 }
887 }
888 else
889 s->status = BUSY_STATE;
890 }
891 #endif
892
893 /* Flush as much pending output as possible */
894 if (s->pending != 0) {
895 flush_pending(strm);
896 if (strm->avail_out == 0) {
897 /* Since avail_out is 0, deflate will be called again with
898 * more output space, but possibly with both pending and
899 * avail_in equal to zero. There won't be anything to do,
900 * but this is not an error situation so make sure we
901 * return OK instead of BUF_ERROR at next call of deflate:
902 */
903 s->last_flush = -1;
904 return Z_OK;
905 }
906
907 /* Make sure there is something to do and avoid duplicate consecutive
908 * flushes. For repeated and useless calls with Z_FINISH, we keep
909 * returning Z_STREAM_END instead of Z_BUF_ERROR.
910 */
911 } else if (strm->avail_in == 0 && RANK(flush) <= RANK(old_flush) &&
912 flush != Z_FINISH) {
913 ERR_RETURN(strm, Z_BUF_ERROR);
914 }
915
916 /* User must not provide more input after the first FINISH: */
917 if (s->status == FINISH_STATE && strm->avail_in != 0) {
918 ERR_RETURN(strm, Z_BUF_ERROR);
919 }
920
921 /* Start a new block or continue the current one.
922 */
923 if (strm->avail_in != 0 || s->lookahead != 0 ||
924 (flush != Z_NO_FLUSH && s->status != FINISH_STATE)) {
925 block_state bstate;
926
927 bstate = s->strategy == Z_HUFFMAN_ONLY ? deflate_huff(s, flush) :
928 (s->strategy == Z_RLE ? deflate_rle(s, flush) :
929 (*(configuration_table[s->level].func))(s, flush));
930
931 if (bstate == finish_started || bstate == finish_done) {
932 s->status = FINISH_STATE;
933 }
934 if (bstate == need_more || bstate == finish_started) {
935 if (strm->avail_out == 0) {
936 s->last_flush = -1; /* avoid BUF_ERROR next call, see above */
937 }
938 return Z_OK;
939 /* If flush != Z_NO_FLUSH && avail_out == 0, the next call
940 * of deflate should use the same flush parameter to make sure
941 * that the flush is complete. So we don't have to output an
942 * empty block here, this will be done at next call. This also
943 * ensures that for a very small output buffer, we emit at most
944 * one empty block.
945 */
946 }
947 if (bstate == block_done) {
948 if (flush == Z_PARTIAL_FLUSH) {
949 _tr_align(s);
950 } else if (flush != Z_BLOCK) { /* FULL_FLUSH or SYNC_FLUSH */
951 _tr_stored_block(s, (char*)0, 0L, 0);
952 /* For a full flush, this empty block will be recognized
953 * as a special marker by inflate_sync().
954 */
955 if (flush == Z_FULL_FLUSH) {
956 CLEAR_HASH(s); /* forget history */
957 if (s->lookahead == 0) {
958 s->strstart = 0;
959 s->block_start = 0L;
960 s->insert = 0;
961 }
962 }
963 }
964 flush_pending(strm);
965 if (strm->avail_out == 0) {
966 s->last_flush = -1; /* avoid BUF_ERROR at next call, see above */
967 return Z_OK;
968 }
969 }
970 }
971 Assert(strm->avail_out > 0, "bug2");
972
973 if (flush != Z_FINISH) return Z_OK;
974 if (s->wrap <= 0) return Z_STREAM_END;
975
976 /* Write the trailer */
977 #ifdef GZIP
978 if (s->wrap == 2) {
979 put_byte(s, (Byte)(strm->adler & 0xff));
980 put_byte(s, (Byte)((strm->adler >> 8) & 0xff));
981 put_byte(s, (Byte)((strm->adler >> 16) & 0xff));
982 put_byte(s, (Byte)((strm->adler >> 24) & 0xff));
983 put_byte(s, (Byte)(strm->total_in & 0xff));
984 put_byte(s, (Byte)((strm->total_in >> 8) & 0xff));
985 put_byte(s, (Byte)((strm->total_in >> 16) & 0xff));
986 put_byte(s, (Byte)((strm->total_in >> 24) & 0xff));
987 }
988 else
989 #endif
990 {
991 putShortMSB(s, (uInt)(strm->adler >> 16));
992 putShortMSB(s, (uInt)(strm->adler & 0xffff));
993 }
994 flush_pending(strm);
995 /* If avail_out is zero, the application will call deflate again
996 * to flush the rest.
997 */
998 if (s->wrap > 0) s->wrap = -s->wrap; /* write the trailer only once! */
999 return s->pending != 0 ? Z_OK : Z_STREAM_END;
1000 }
1001
1002 /* ========================================================================= */
1003 int ZEXPORT deflateEnd (strm)
1004 z_streamp strm;
1005 {
1006 int status;
1007
1008 if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR;
1009
1010 status = strm->state->status;
1011 if (status != INIT_STATE &&
1012 status != EXTRA_STATE &&
1013 status != NAME_STATE &&
1014 status != COMMENT_STATE &&
1015 status != HCRC_STATE &&
1016 status != BUSY_STATE &&
1017 status != FINISH_STATE) {
1018 return Z_STREAM_ERROR;
1019 }
1020
1021 /* Deallocate in reverse order of allocations: */
1022 TRY_FREE(strm, strm->state->pending_buf);
1023 TRY_FREE(strm, strm->state->head);
1024 TRY_FREE(strm, strm->state->prev);
1025 TRY_FREE(strm, strm->state->window);
1026
1027 ZFREE(strm, strm->state);
1028 strm->state = Z_NULL;
1029
1030 return status == BUSY_STATE ? Z_DATA_ERROR : Z_OK;
1031 }
1032
1033 /* =========================================================================
1034 * Copy the source state to the destination state.
1035 * To simplify the source, this is not supported for 16-bit MSDOS (which
1036 * doesn't have enough memory anyway to duplicate compression states).
1037 */
1038 int ZEXPORT deflateCopy (dest, source)
1039 z_streamp dest;
1040 z_streamp source;
1041 {
1042 #ifdef MAXSEG_64K
1043 return Z_STREAM_ERROR;
1044 #else
1045 deflate_state *ds;
1046 deflate_state *ss;
1047 ushf *overlay;
1048
1049
1050 if (source == Z_NULL || dest == Z_NULL || source->state == Z_NULL) {
1051 return Z_STREAM_ERROR;
1052 }
1053
1054 ss = source->state;
1055
1056 zmemcpy((voidpf)dest, (voidpf)source, sizeof(z_stream));
1057
1058 ds = (deflate_state *) ZALLOC(dest, 1, sizeof(deflate_state));
1059 if (ds == Z_NULL) return Z_MEM_ERROR;
1060 dest->state = (struct internal_state FAR *) ds;
1061 zmemcpy((voidpf)ds, (voidpf)ss, sizeof(deflate_state));
1062 ds->strm = dest;
1063
1064 ds->window = (Bytef *) ZALLOC(dest, ds->w_size, 2*sizeof(Byte));
1065 ds->prev = (Posf *) ZALLOC(dest, ds->w_size, sizeof(Pos));
1066 ds->head = (Posf *) ZALLOC(dest, ds->hash_size, sizeof(Pos));
1067 overlay = (ushf *) ZALLOC(dest, ds->lit_bufsize, sizeof(ush)+2);
1068 ds->pending_buf = (uchf *) overlay;
1069
1070 if (ds->window == Z_NULL || ds->prev == Z_NULL || ds->head == Z_NULL ||
1071 ds->pending_buf == Z_NULL) {
1072 deflateEnd (dest);
1073 return Z_MEM_ERROR;
1074 }
1075 /* following zmemcpy do not work for 16-bit MSDOS */
1076 zmemcpy(ds->window, ss->window, ds->w_size * 2 * sizeof(Byte));
1077 zmemcpy((voidpf)ds->prev, (voidpf)ss->prev, ds->w_size * sizeof(Pos));
1078 zmemcpy((voidpf)ds->head, (voidpf)ss->head, ds->hash_size * sizeof(Pos));
1079 zmemcpy(ds->pending_buf, ss->pending_buf, (uInt)ds->pending_buf_size);
1080
1081 ds->pending_out = ds->pending_buf + (ss->pending_out - ss->pending_buf);
1082 ds->d_buf = overlay + ds->lit_bufsize/sizeof(ush);
1083 ds->l_buf = ds->pending_buf + (1+sizeof(ush))*ds->lit_bufsize;
1084
1085 ds->l_desc.dyn_tree = ds->dyn_ltree;
1086 ds->d_desc.dyn_tree = ds->dyn_dtree;
1087 ds->bl_desc.dyn_tree = ds->bl_tree;
1088
1089 return Z_OK;
1090 #endif /* MAXSEG_64K */
1091 }
1092
1093 /* ===========================================================================
1094 * Read a new buffer from the current input stream, update the adler32
1095 * and total number of bytes read. All deflate() input goes through
1096 * this function so some applications may wish to modify it to avoid
1097 * allocating a large strm->next_in buffer and copying from it.
1098 * (See also flush_pending()).
1099 */
1100 local int read_buf(strm, buf, size)
1101 z_streamp strm;
1102 Bytef *buf;
1103 unsigned size;
1104 {
1105 unsigned len = strm->avail_in;
1106
1107 if (len > size) len = size;
1108 if (len == 0) return 0;
1109
1110 strm->avail_in -= len;
1111
1112 zmemcpy(buf, strm->next_in, len);
1113 if (strm->state->wrap == 1) {
1114 strm->adler = adler32(strm->adler, buf, len);
1115 }
1116 #ifdef GZIP
1117 else if (strm->state->wrap == 2) {
1118 strm->adler = crc32(strm->adler, buf, len);
1119 }
1120 #endif
1121 strm->next_in += len;
1122 strm->total_in += len;
1123
1124 return (int)len;
1125 }
1126
1127 /* ===========================================================================
1128 * Initialize the "longest match" routines for a new zlib stream
1129 */
1130 local void lm_init (s)
1131 deflate_state *s;
1132 {
1133 s->window_size = (ulg)2L*s->w_size;
1134
1135 CLEAR_HASH(s);
1136
1137 /* Set the default configuration parameters:
1138 */
1139 s->max_lazy_match = configuration_table[s->level].max_lazy;
1140 s->good_match = configuration_table[s->level].good_length;
1141 s->nice_match = configuration_table[s->level].nice_length;
1142 s->max_chain_length = configuration_table[s->level].max_chain;
1143
1144 s->strstart = 0;
1145 s->block_start = 0L;
1146 s->lookahead = 0;
1147 s->insert = 0;
1148 s->match_length = s->prev_length = MIN_MATCH-1;
1149 s->match_available = 0;
1150 s->ins_h = 0;
1151 #ifndef FASTEST
1152 #ifdef ASMV
1153 match_init(); /* initialize the asm code */
1154 #endif
1155 #endif
1156 }
1157
1158 #ifndef FASTEST
1159 /* ===========================================================================
1160 * Set match_start to the longest match starting at the given string and
1161 * return its length. Matches shorter or equal to prev_length are discarded,
1162 * in which case the result is equal to prev_length and match_start is
1163 * garbage.
1164 * IN assertions: cur_match is the head of the hash chain for the current
1165 * string (strstart) and its distance is <= MAX_DIST, and prev_length >= 1
1166 * OUT assertion: the match length is not greater than s->lookahead.
1167 */
1168 #ifndef ASMV
1169 /* For 80x86 and 680x0, an optimized version will be provided in match.asm or
1170 * match.S. The code will be functionally equivalent.
1171 */
1172 local uInt longest_match(s, cur_match)
1173 deflate_state *s;
1174 IPos cur_match; /* current match */
1175 {
1176 unsigned chain_length = s->max_chain_length;/* max hash chain length */
1177 register Bytef *scan = s->window + s->strstart; /* current string */
1178 register Bytef *match; /* matched string */
1179 register int len; /* length of current match */
1180 int best_len = s->prev_length; /* best match length so far */
1181 int nice_match = s->nice_match; /* stop if match long enough */
1182 IPos limit = s->strstart > (IPos)MAX_DIST(s) ?
1183 s->strstart - (IPos)MAX_DIST(s) : NIL;
1184 /* Stop when cur_match becomes <= limit. To simplify the code,
1185 * we prevent matches with the string of window index 0.
1186 */
1187 Posf *prev = s->prev;
1188 uInt wmask = s->w_mask;
1189
1190 #ifdef UNALIGNED_OK
1191 /* Compare two bytes at a time. Note: this is not always beneficial.
1192 * Try with and without -DUNALIGNED_OK to check.
1193 */
1194 register Bytef *strend = s->window + s->strstart + MAX_MATCH - 1;
1195 register ush scan_start = *(ushf*)scan;
1196 register ush scan_end = *(ushf*)(scan+best_len-1);
1197 #else
1198 register Bytef *strend = s->window + s->strstart + MAX_MATCH;
1199 register Byte scan_end1 = scan[best_len-1];
1200 register Byte scan_end = scan[best_len];
1201 #endif
1202
1203 /* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16.
1204 * It is easy to get rid of this optimization if necessary.
1205 */
1206 Assert(s->hash_bits >= 8 && MAX_MATCH == 258, "Code too clever");
1207
1208 /* Do not waste too much time if we already have a good match: */
1209 if (s->prev_length >= s->good_match) {
1210 chain_length >>= 2;
1211 }
1212 /* Do not look for matches beyond the end of the input. This is necessary
1213 * to make deflate deterministic.
1214 */
1215 if ((uInt)nice_match > s->lookahead) nice_match = s->lookahead;
1216
1217 Assert((ulg)s->strstart <= s->window_size-MIN_LOOKAHEAD, "need lookahead");
1218
1219 do {
1220 Assert(cur_match < s->strstart, "no future");
1221 match = s->window + cur_match;
1222
1223 /* Skip to next match if the match length cannot increase
1224 * or if the match length is less than 2. Note that the checks below
1225 * for insufficient lookahead only occur occasionally for performance
1226 * reasons. Therefore uninitialized memory will be accessed, and
1227 * conditional jumps will be made that depend on those values.
1228 * However the length of the match is limited to the lookahead, so
1229 * the output of deflate is not affected by the uninitialized values.
1230 */
1231 #if (defined(UNALIGNED_OK) && MAX_MATCH == 258)
1232 /* This code assumes sizeof(unsigned short) == 2. Do not use
1233 * UNALIGNED_OK if your compiler uses a different size.
1234 */
1235 if (*(ushf*)(match+best_len-1) != scan_end ||
1236 *(ushf*)match != scan_start) continue;
1237
1238 /* It is not necessary to compare scan[2] and match[2] since they are
1239 * always equal when the other bytes match, given that the hash keys
1240 * are equal and that HASH_BITS >= 8. Compare 2 bytes at a time at
1241 * strstart+3, +5, ... up to strstart+257. We check for insufficient
1242 * lookahead only every 4th comparison; the 128th check will be made
1243 * at strstart+257. If MAX_MATCH-2 is not a multiple of 8, it is
1244 * necessary to put more guard bytes at the end of the window, or
1245 * to check more often for insufficient lookahead.
1246 */
1247 Assert(scan[2] == match[2], "scan[2]?");
1248 scan++, match++;
1249 do {
1250 } while (*(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
1251 *(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
1252 *(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
1253 *(ushf*)(scan+=2) == *(ushf*)(match+=2) &&
1254 scan < strend);
1255 /* The funny "do {}" generates better code on most compilers */
1256
1257 /* Here, scan <= window+strstart+257 */
1258 Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan");
1259 if (*scan == *match) scan++;
1260
1261 len = (MAX_MATCH - 1) - (int)(strend-scan);
1262 scan = strend - (MAX_MATCH-1);
1263
1264 #else /* UNALIGNED_OK */
1265
1266 if (match[best_len] != scan_end ||
1267 match[best_len-1] != scan_end1 ||
1268 *match != *scan ||
1269 *++match != scan[1]) continue;
1270
1271 /* The check at best_len-1 can be removed because it will be made
1272 * again later. (This heuristic is not always a win.)
1273 * It is not necessary to compare scan[2] and match[2] since they
1274 * are always equal when the other bytes match, given that
1275 * the hash keys are equal and that HASH_BITS >= 8.
1276 */
1277 scan += 2, match++;
1278 Assert(*scan == *match, "match[2]?");
1279
1280 /* We check for insufficient lookahead only every 8th comparison;
1281 * the 256th check will be made at strstart+258.
1282 */
1283 do {
1284 } while (*++scan == *++match && *++scan == *++match &&
1285 *++scan == *++match && *++scan == *++match &&
1286 *++scan == *++match && *++scan == *++match &&
1287 *++scan == *++match && *++scan == *++match &&
1288 scan < strend);
1289
1290 Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan");
1291
1292 len = MAX_MATCH - (int)(strend - scan);
1293 scan = strend - MAX_MATCH;
1294
1295 #endif /* UNALIGNED_OK */
1296
1297 if (len > best_len) {
1298 s->match_start = cur_match;
1299 best_len = len;
1300 if (len >= nice_match) break;
1301 #ifdef UNALIGNED_OK
1302 scan_end = *(ushf*)(scan+best_len-1);
1303 #else
1304 scan_end1 = scan[best_len-1];
1305 scan_end = scan[best_len];
1306 #endif
1307 }
1308 } while ((cur_match = prev[cur_match & wmask]) > limit
1309 && --chain_length != 0);
1310
1311 if ((uInt)best_len <= s->lookahead) return (uInt)best_len;
1312 return s->lookahead;
1313 }
1314 #endif /* ASMV */
1315
1316 #else /* FASTEST */
1317
1318 /* ---------------------------------------------------------------------------
1319 * Optimized version for FASTEST only
1320 */
1321 local uInt longest_match(s, cur_match)
1322 deflate_state *s;
1323 IPos cur_match; /* current match */
1324 {
1325 register Bytef *scan = s->window + s->strstart; /* current string */
1326 register Bytef *match; /* matched string */
1327 register int len; /* length of current match */
1328 register Bytef *strend = s->window + s->strstart + MAX_MATCH;
1329
1330 /* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16.
1331 * It is easy to get rid of this optimization if necessary.
1332 */
1333 Assert(s->hash_bits >= 8 && MAX_MATCH == 258, "Code too clever");
1334
1335 Assert((ulg)s->strstart <= s->window_size-MIN_LOOKAHEAD, "need lookahead");
1336
1337 Assert(cur_match < s->strstart, "no future");
1338
1339 match = s->window + cur_match;
1340
1341 /* Return failure if the match length is less than 2:
1342 */
1343 if (match[0] != scan[0] || match[1] != scan[1]) return MIN_MATCH-1;
1344
1345 /* The check at best_len-1 can be removed because it will be made
1346 * again later. (This heuristic is not always a win.)
1347 * It is not necessary to compare scan[2] and match[2] since they
1348 * are always equal when the other bytes match, given that
1349 * the hash keys are equal and that HASH_BITS >= 8.
1350 */
1351 scan += 2, match += 2;
1352 Assert(*scan == *match, "match[2]?");
1353
1354 /* We check for insufficient lookahead only every 8th comparison;
1355 * the 256th check will be made at strstart+258.
1356 */
1357 do {
1358 } while (*++scan == *++match && *++scan == *++match &&
1359 *++scan == *++match && *++scan == *++match &&
1360 *++scan == *++match && *++scan == *++match &&
1361 *++scan == *++match && *++scan == *++match &&
1362 scan < strend);
1363
1364 Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan");
1365
1366 len = MAX_MATCH - (int)(strend - scan);
1367
1368 if (len < MIN_MATCH) return MIN_MATCH - 1;
1369
1370 s->match_start = cur_match;
1371 return (uInt)len <= s->lookahead ? (uInt)len : s->lookahead;
1372 }
1373
1374 #endif /* FASTEST */
1375
1376 #ifdef DEBUG
1377 /* ===========================================================================
1378 * Check that the match at match_start is indeed a match.
1379 */
1380 local void check_match(s, start, match, length)
1381 deflate_state *s;
1382 IPos start, match;
1383 int length;
1384 {
1385 /* check that the match is indeed a match */
1386 if (zmemcmp(s->window + match,
1387 s->window + start, length) != EQUAL) {
1388 fprintf(stderr, " start %u, match %u, length %d\n",
1389 start, match, length);
1390 do {
1391 fprintf(stderr, "%c%c", s->window[match++], s->window[start++]);
1392 } while (--length != 0);
1393 z_error("invalid match");
1394 }
1395 if (z_verbose > 1) {
1396 fprintf(stderr,"\\[%d,%d]", start-match, length);
1397 do { putc(s->window[start++], stderr); } while (--length != 0);
1398 }
1399 }
1400 #else
1401 # define check_match(s, start, match, length)
1402 #endif /* DEBUG */
1403
1404 /* ===========================================================================
1405 * Fill the window when the lookahead becomes insufficient.
1406 * Updates strstart and lookahead.
1407 *
1408 * IN assertion: lookahead < MIN_LOOKAHEAD
1409 * OUT assertions: strstart <= window_size-MIN_LOOKAHEAD
1410 * At least one byte has been read, or avail_in == 0; reads are
1411 * performed for at least two bytes (required for the zip translate_eol
1412 * option -- not supported here).
1413 */
1414 local void fill_window(s)
1415 deflate_state *s;
1416 {
1417 register unsigned n, m;
1418 register Posf *p;
1419 unsigned more; /* Amount of free space at the end of the window. */
1420 uInt wsize = s->w_size;
1421
1422 Assert(s->lookahead < MIN_LOOKAHEAD, "already enough lookahead");
1423
1424 do {
1425 more = (unsigned)(s->window_size -(ulg)s->lookahead -(ulg)s->strstart);
1426
1427 /* Deal with !@#$% 64K limit: */
1428 if (sizeof(int) <= 2) {
1429 if (more == 0 && s->strstart == 0 && s->lookahead == 0) {
1430 more = wsize;
1431
1432 } else if (more == (unsigned)(-1)) {
1433 /* Very unlikely, but possible on 16 bit machine if
1434 * strstart == 0 && lookahead == 1 (input done a byte at time)
1435 */
1436 more--;
1437 }
1438 }
1439
1440 /* If the window is almost full and there is insufficient lookahead,
1441 * move the upper half to the lower one to make room in the upper half.
1442 */
1443 if (s->strstart >= wsize+MAX_DIST(s)) {
1444
1445 zmemcpy(s->window, s->window+wsize, (unsigned)wsize);
1446 s->match_start -= wsize;
1447 s->strstart -= wsize; /* we now have strstart >= MAX_DIST */
1448 s->block_start -= (long) wsize;
1449
1450 /* Slide the hash table (could be avoided with 32 bit values
1451 at the expense of memory usage). We slide even when level == 0
1452 to keep the hash table consistent if we switch back to level > 0
1453 later. (Using level 0 permanently is not an optimal usage of
1454 zlib, so we don't care about this pathological case.)
1455 */
1456 n = s->hash_size;
1457 p = &s->head[n];
1458 do {
1459 m = *--p;
1460 *p = (Pos)(m >= wsize ? m-wsize : NIL);
1461 } while (--n);
1462
1463 n = wsize;
1464 #ifndef FASTEST
1465 p = &s->prev[n];
1466 do {
1467 m = *--p;
1468 *p = (Pos)(m >= wsize ? m-wsize : NIL);
1469 /* If n is not on any hash chain, prev[n] is garbage but
1470 * its value will never be used.
1471 */
1472 } while (--n);
1473 #endif
1474 more += wsize;
1475 }
1476 if (s->strm->avail_in == 0) break;
1477
1478 /* If there was no sliding:
1479 * strstart <= WSIZE+MAX_DIST-1 && lookahead <= MIN_LOOKAHEAD - 1 &&
1480 * more == window_size - lookahead - strstart
1481 * => more >= window_size - (MIN_LOOKAHEAD-1 + WSIZE + MAX_DIST-1)
1482 * => more >= window_size - 2*WSIZE + 2
1483 * In the BIG_MEM or MMAP case (not yet supported),
1484 * window_size == input_size + MIN_LOOKAHEAD &&
1485 * strstart + s->lookahead <= input_size => more >= MIN_LOOKAHEAD.
1486 * Otherwise, window_size == 2*WSIZE so more >= 2.
1487 * If there was sliding, more >= WSIZE. So in all cases, more >= 2.
1488 */
1489 Assert(more >= 2, "more < 2");
1490
1491 n = read_buf(s->strm, s->window + s->strstart + s->lookahead, more);
1492 s->lookahead += n;
1493
1494 /* Initialize the hash value now that we have some input: */
1495 if (s->lookahead + s->insert >= MIN_MATCH) {
1496 uInt str = s->strstart - s->insert;
1497 s->ins_h = s->window[str];
1498 UPDATE_HASH(s, s->ins_h, s->window[str + 1]);
1499 #if MIN_MATCH != 3
1500 Call UPDATE_HASH() MIN_MATCH-3 more times
1501 #endif
1502 while (s->insert) {
1503 UPDATE_HASH(s, s->ins_h, s->window[str + MIN_MATCH-1]);
1504 #ifndef FASTEST
1505 s->prev[str & s->w_mask] = s->head[s->ins_h];
1506 #endif
1507 s->head[s->ins_h] = (Pos)str;
1508 str++;
1509 s->insert--;
1510 if (s->lookahead + s->insert < MIN_MATCH)
1511 break;
1512 }
1513 }
1514 /* If the whole input has less than MIN_MATCH bytes, ins_h is garbage,
1515 * but this is not important since only literal bytes will be emitted.
1516 */
1517
1518 } while (s->lookahead < MIN_LOOKAHEAD && s->strm->avail_in != 0);
1519
1520 /* If the WIN_INIT bytes after the end of the current data have never been
1521 * written, then zero those bytes in order to avoid memory check reports of
1522 * the use of uninitialized (or uninitialised as Julian writes) bytes by
1523 * the longest match routines. Update the high water mark for the next
1524 * time through here. WIN_INIT is set to MAX_MATCH since the longest match
1525 * routines allow scanning to strstart + MAX_MATCH, ignoring lookahead.
1526 */
1527 if (s->high_water < s->window_size) {
1528 ulg curr = s->strstart + (ulg)(s->lookahead);
1529 ulg init;
1530
1531 if (s->high_water < curr) {
1532 /* Previous high water mark below current data -- zero WIN_INIT
1533 * bytes or up to end of window, whichever is less.
1534 */
1535 init = s->window_size - curr;
1536 if (init > WIN_INIT)
1537 init = WIN_INIT;
1538 zmemzero(s->window + curr, (unsigned)init);
1539 s->high_water = curr + init;
1540 }
1541 else if (s->high_water < (ulg)curr + WIN_INIT) {
1542 /* High water mark at or above current data, but below current data
1543 * plus WIN_INIT -- zero out to current data plus WIN_INIT, or up
1544 * to end of window, whichever is less.
1545 */
1546 init = (ulg)curr + WIN_INIT - s->high_water;
1547 if (init > s->window_size - s->high_water)
1548 init = s->window_size - s->high_water;
1549 zmemzero(s->window + s->high_water, (unsigned)init);
1550 s->high_water += init;
1551 }
1552 }
1553
1554 Assert((ulg)s->strstart <= s->window_size - MIN_LOOKAHEAD,
1555 "not enough room for search");
1556 }
1557
1558 /* ===========================================================================
1559 * Flush the current block, with given end-of-file flag.
1560 * IN assertion: strstart is set to the end of the current match.
1561 */
1562 #define FLUSH_BLOCK_ONLY(s, last) { \
1563 _tr_flush_block(s, (s->block_start >= 0L ? \
1564 (charf *)&s->window[(unsigned)s->block_start] : \
1565 (charf *)Z_NULL), \
1566 (ulg)((long)s->strstart - s->block_start), \
1567 (last)); \
1568 s->block_start = s->strstart; \
1569 flush_pending(s->strm); \
1570 Tracev((stderr,"[FLUSH]")); \
1571 }
1572
1573 /* Same but force premature exit if necessary. */
1574 #define FLUSH_BLOCK(s, last) { \
1575 FLUSH_BLOCK_ONLY(s, last); \
1576 if (s->strm->avail_out == 0) return (last) ? finish_started : need_more; \
1577 }
1578
1579 /* ===========================================================================
1580 * Copy without compression as much as possible from the input stream, return
1581 * the current block state.
1582 * This function does not insert new strings in the dictionary since
1583 * uncompressible data is probably not useful. This function is used
1584 * only for the level=0 compression option.
1585 * NOTE: this function should be optimized to avoid extra copying from
1586 * window to pending_buf.
1587 */
1588 local block_state deflate_stored(s, flush)
1589 deflate_state *s;
1590 int flush;
1591 {
1592 /* Stored blocks are limited to 0xffff bytes, pending_buf is limited
1593 * to pending_buf_size, and each stored block has a 5 byte header:
1594 */
1595 ulg max_block_size = 0xffff;
1596 ulg max_start;
1597
1598 if (max_block_size > s->pending_buf_size - 5) {
1599 max_block_size = s->pending_buf_size - 5;
1600 }
1601
1602 /* Copy as much as possible from input to output: */
1603 for (;;) {
1604 /* Fill the window as much as possible: */
1605 if (s->lookahead <= 1) {
1606
1607 Assert(s->strstart < s->w_size+MAX_DIST(s) ||
1608 s->block_start >= (long)s->w_size, "slide too late");
1609
1610 fill_window(s);
1611 if (s->lookahead == 0 && flush == Z_NO_FLUSH) return need_more;
1612
1613 if (s->lookahead == 0) break; /* flush the current block */
1614 }
1615 Assert(s->block_start >= 0L, "block gone");
1616
1617 s->strstart += s->lookahead;
1618 s->lookahead = 0;
1619
1620 /* Emit a stored block if pending_buf will be full: */
1621 max_start = s->block_start + max_block_size;
1622 if (s->strstart == 0 || (ulg)s->strstart >= max_start) {
1623 /* strstart == 0 is possible when wraparound on 16-bit machine */
1624 s->lookahead = (uInt)(s->strstart - max_start);
1625 s->strstart = (uInt)max_start;
1626 FLUSH_BLOCK(s, 0);
1627 }
1628 /* Flush if we may have to slide, otherwise block_start may become
1629 * negative and the data will be gone:
1630 */
1631 if (s->strstart - (uInt)s->block_start >= MAX_DIST(s)) {
1632 FLUSH_BLOCK(s, 0);
1633 }
1634 }
1635 s->insert = 0;
1636 if (flush == Z_FINISH) {
1637 FLUSH_BLOCK(s, 1);
1638 return finish_done;
1639 }
1640 if ((long)s->strstart > s->block_start)
1641 FLUSH_BLOCK(s, 0);
1642 return block_done;
1643 }
1644
1645 /* ===========================================================================
1646 * Compress as much as possible from the input stream, return the current
1647 * block state.
1648 * This function does not perform lazy evaluation of matches and inserts
1649 * new strings in the dictionary only for unmatched strings or for short
1650 * matches. It is used only for the fast compression options.
1651 */
1652 local block_state deflate_fast(s, flush)
1653 deflate_state *s;
1654 int flush;
1655 {
1656 IPos hash_head; /* head of the hash chain */
1657 int bflush; /* set if current block must be flushed */
1658
1659 for (;;) {
1660 /* Make sure that we always have enough lookahead, except
1661 * at the end of the input file. We need MAX_MATCH bytes
1662 * for the next match, plus MIN_MATCH bytes to insert the
1663 * string following the next match.
1664 */
1665 if (s->lookahead < MIN_LOOKAHEAD) {
1666 fill_window(s);
1667 if (s->lookahead < MIN_LOOKAHEAD && flush == Z_NO_FLUSH) {
1668 return need_more;
1669 }
1670 if (s->lookahead == 0) break; /* flush the current block */
1671 }
1672
1673 /* Insert the string window[strstart .. strstart+2] in the
1674 * dictionary, and set hash_head to the head of the hash chain:
1675 */
1676 hash_head = NIL;
1677 if (s->lookahead >= MIN_MATCH) {
1678 INSERT_STRING(s, s->strstart, hash_head);
1679 }
1680
1681 /* Find the longest match, discarding those <= prev_length.
1682 * At this point we have always match_length < MIN_MATCH
1683 */
1684 if (hash_head != NIL && s->strstart - hash_head <= MAX_DIST(s)) {
1685 /* To simplify the code, we prevent matches with the string
1686 * of window index 0 (in particular we have to avoid a match
1687 * of the string with itself at the start of the input file).
1688 */
1689 s->match_length = longest_match (s, hash_head);
1690 /* longest_match() sets match_start */
1691 }
1692 if (s->match_length >= MIN_MATCH) {
1693 check_match(s, s->strstart, s->match_start, s->match_length);
1694
1695 _tr_tally_dist(s, s->strstart - s->match_start,
1696 s->match_length - MIN_MATCH, bflush);
1697
1698 s->lookahead -= s->match_length;
1699
1700 /* Insert new strings in the hash table only if the match length
1701 * is not too large. This saves time but degrades compression.
1702 */
1703 #ifndef FASTEST
1704 if (s->match_length <= s->max_insert_length &&
1705 s->lookahead >= MIN_MATCH) {
1706 s->match_length--; /* string at strstart already in table */
1707 do {
1708 s->strstart++;
1709 INSERT_STRING(s, s->strstart, hash_head);
1710 /* strstart never exceeds WSIZE-MAX_MATCH, so there are
1711 * always MIN_MATCH bytes ahead.
1712 */
1713 } while (--s->match_length != 0);
1714 s->strstart++;
1715 } else
1716 #endif
1717 {
1718 s->strstart += s->match_length;
1719 s->match_length = 0;
1720 s->ins_h = s->window[s->strstart];
1721 UPDATE_HASH(s, s->ins_h, s->window[s->strstart+1]);
1722 #if MIN_MATCH != 3
1723 Call UPDATE_HASH() MIN_MATCH-3 more times
1724 #endif
1725 /* If lookahead < MIN_MATCH, ins_h is garbage, but it does not
1726 * matter since it will be recomputed at next deflate call.
1727 */
1728 }
1729 } else {
1730 /* No match, output a literal byte */
1731 Tracevv((stderr,"%c", s->window[s->strstart]));
1732 _tr_tally_lit (s, s->window[s->strstart], bflush);
1733 s->lookahead--;
1734 s->strstart++;
1735 }
1736 if (bflush) FLUSH_BLOCK(s, 0);
1737 }
1738 s->insert = s->strstart < MIN_MATCH-1 ? s->strstart : MIN_MATCH-1;
1739 if (flush == Z_FINISH) {
1740 FLUSH_BLOCK(s, 1);
1741 return finish_done;
1742 }
1743 if (s->last_lit)
1744 FLUSH_BLOCK(s, 0);
1745 return block_done;
1746 }
1747
1748 #ifndef FASTEST
1749 /* ===========================================================================
1750 * Same as above, but achieves better compression. We use a lazy
1751 * evaluation for matches: a match is finally adopted only if there is
1752 * no better match at the next window position.
1753 */
1754 local block_state deflate_slow(s, flush)
1755 deflate_state *s;
1756 int flush;
1757 {
1758 IPos hash_head; /* head of hash chain */
1759 int bflush; /* set if current block must be flushed */
1760
1761 /* Process the input block. */
1762 for (;;) {
1763 /* Make sure that we always have enough lookahead, except
1764 * at the end of the input file. We need MAX_MATCH bytes
1765 * for the next match, plus MIN_MATCH bytes to insert the
1766 * string following the next match.
1767 */
1768 if (s->lookahead < MIN_LOOKAHEAD) {
1769 fill_window(s);
1770 if (s->lookahead < MIN_LOOKAHEAD && flush == Z_NO_FLUSH) {
1771 return need_more;
1772 }
1773 if (s->lookahead == 0) break; /* flush the current block */
1774 }
1775
1776 /* Insert the string window[strstart .. strstart+2] in the
1777 * dictionary, and set hash_head to the head of the hash chain:
1778 */
1779 hash_head = NIL;
1780 if (s->lookahead >= MIN_MATCH) {
1781 INSERT_STRING(s, s->strstart, hash_head);
1782 }
1783
1784 /* Find the longest match, discarding those <= prev_length.
1785 */
1786 s->prev_length = s->match_length, s->prev_match = s->match_start;
1787 s->match_length = MIN_MATCH-1;
1788
1789 if (hash_head != NIL && s->prev_length < s->max_lazy_match &&
1790 s->strstart - hash_head <= MAX_DIST(s)) {
1791 /* To simplify the code, we prevent matches with the string
1792 * of window index 0 (in particular we have to avoid a match
1793 * of the string with itself at the start of the input file).
1794 */
1795 s->match_length = longest_match (s, hash_head);
1796 /* longest_match() sets match_start */
1797
1798 if (s->match_length <= 5 && (s->strategy == Z_FILTERED
1799 #if TOO_FAR <= 32767
1800 || (s->match_length == MIN_MATCH &&
1801 s->strstart - s->match_start > TOO_FAR)
1802 #endif
1803 )) {
1804
1805 /* If prev_match is also MIN_MATCH, match_start is garbage
1806 * but we will ignore the current match anyway.
1807 */
1808 s->match_length = MIN_MATCH-1;
1809 }
1810 }
1811 /* If there was a match at the previous step and the current
1812 * match is not better, output the previous match:
1813 */
1814 if (s->prev_length >= MIN_MATCH && s->match_length <= s->prev_length) {
1815 uInt max_insert = s->strstart + s->lookahead - MIN_MATCH;
1816 /* Do not insert strings in hash table beyond this. */
1817
1818 check_match(s, s->strstart-1, s->prev_match, s->prev_length);
1819
1820 _tr_tally_dist(s, s->strstart -1 - s->prev_match,
1821 s->prev_length - MIN_MATCH, bflush);
1822
1823 /* Insert in hash table all strings up to the end of the match.
1824 * strstart-1 and strstart are already inserted. If there is not
1825 * enough lookahead, the last two strings are not inserted in
1826 * the hash table.
1827 */
1828 s->lookahead -= s->prev_length-1;
1829 s->prev_length -= 2;
1830 do {
1831 if (++s->strstart <= max_insert) {
1832 INSERT_STRING(s, s->strstart, hash_head);
1833 }
1834 } while (--s->prev_length != 0);
1835 s->match_available = 0;
1836 s->match_length = MIN_MATCH-1;
1837 s->strstart++;
1838
1839 if (bflush) FLUSH_BLOCK(s, 0);
1840
1841 } else if (s->match_available) {
1842 /* If there was no match at the previous position, output a
1843 * single literal. If there was a match but the current match
1844 * is longer, truncate the previous match to a single literal.
1845 */
1846 Tracevv((stderr,"%c", s->window[s->strstart-1]));
1847 _tr_tally_lit(s, s->window[s->strstart-1], bflush);
1848 if (bflush) {
1849 FLUSH_BLOCK_ONLY(s, 0);
1850 }
1851 s->strstart++;
1852 s->lookahead--;
1853 if (s->strm->avail_out == 0) return need_more;
1854 } else {
1855 /* There is no previous match to compare with, wait for
1856 * the next step to decide.
1857 */
1858 s->match_available = 1;
1859 s->strstart++;
1860 s->lookahead--;
1861 }
1862 }
1863 Assert (flush != Z_NO_FLUSH, "no flush?");
1864 if (s->match_available) {
1865 Tracevv((stderr,"%c", s->window[s->strstart-1]));
1866 _tr_tally_lit(s, s->window[s->strstart-1], bflush);
1867 s->match_available = 0;
1868 }
1869 s->insert = s->strstart < MIN_MATCH-1 ? s->strstart : MIN_MATCH-1;
1870 if (flush == Z_FINISH) {
1871 FLUSH_BLOCK(s, 1);
1872 return finish_done;
1873 }
1874 if (s->last_lit)
1875 FLUSH_BLOCK(s, 0);
1876 return block_done;
1877 }
1878 #endif /* FASTEST */
1879
1880 /* ===========================================================================
1881 * For Z_RLE, simply look for runs of bytes, generate matches only of distance
1882 * one. Do not maintain a hash table. (It will be regenerated if this run of
1883 * deflate switches away from Z_RLE.)
1884 */
1885 local block_state deflate_rle(s, flush)
1886 deflate_state *s;
1887 int flush;
1888 {
1889 int bflush; /* set if current block must be flushed */
1890 uInt prev; /* byte at distance one to match */
1891 Bytef *scan, *strend; /* scan goes up to strend for length of run */
1892
1893 for (;;) {
1894 /* Make sure that we always have enough lookahead, except
1895 * at the end of the input file. We need MAX_MATCH bytes
1896 * for the longest run, plus one for the unrolled loop.
1897 */
1898 if (s->lookahead <= MAX_MATCH) {
1899 fill_window(s);
1900 if (s->lookahead <= MAX_MATCH && flush == Z_NO_FLUSH) {
1901 return need_more;
1902 }
1903 if (s->lookahead == 0) break; /* flush the current block */
1904 }
1905
1906 /* See how many times the previous byte repeats */
1907 s->match_length = 0;
1908 if (s->lookahead >= MIN_MATCH && s->strstart > 0) {
1909 scan = s->window + s->strstart - 1;
1910 prev = *scan;
1911 if (prev == *++scan && prev == *++scan && prev == *++scan) {
1912 strend = s->window + s->strstart + MAX_MATCH;
1913 do {
1914 } while (prev == *++scan && prev == *++scan &&
1915 prev == *++scan && prev == *++scan &&
1916 prev == *++scan && prev == *++scan &&
1917 prev == *++scan && prev == *++scan &&
1918 scan < strend);
1919 s->match_length = MAX_MATCH - (int)(strend - scan);
1920 if (s->match_length > s->lookahead)
1921 s->match_length = s->lookahead;
1922 }
1923 Assert(scan <= s->window+(uInt)(s->window_size-1), "wild scan");
1924 }
1925
1926 /* Emit match if have run of MIN_MATCH or longer, else emit literal */
1927 if (s->match_length >= MIN_MATCH) {
1928 check_match(s, s->strstart, s->strstart - 1, s->match_length);
1929
1930 _tr_tally_dist(s, 1, s->match_length - MIN_MATCH, bflush);
1931
1932 s->lookahead -= s->match_length;
1933 s->strstart += s->match_length;
1934 s->match_length = 0;
1935 } else {
1936 /* No match, output a literal byte */
1937 Tracevv((stderr,"%c", s->window[s->strstart]));
1938 _tr_tally_lit (s, s->window[s->strstart], bflush);
1939 s->lookahead--;
1940 s->strstart++;
1941 }
1942 if (bflush) FLUSH_BLOCK(s, 0);
1943 }
1944 s->insert = 0;
1945 if (flush == Z_FINISH) {
1946 FLUSH_BLOCK(s, 1);
1947 return finish_done;
1948 }
1949 if (s->last_lit)
1950 FLUSH_BLOCK(s, 0);
1951 return block_done;
1952 }
1953
1954 /* ===========================================================================
1955 * For Z_HUFFMAN_ONLY, do not look for matches. Do not maintain a hash table.
1956 * (It will be regenerated if this run of deflate switches away from Huffman.)
1957 */
1958 local block_state deflate_huff(s, flush)
1959 deflate_state *s;
1960 int flush;
1961 {
1962 int bflush; /* set if current block must be flushed */
1963
1964 for (;;) {
1965 /* Make sure that we have a literal to write. */
1966 if (s->lookahead == 0) {
1967 fill_window(s);
1968 if (s->lookahead == 0) {
1969 if (flush == Z_NO_FLUSH)
1970 return need_more;
1971 break; /* flush the current block */
1972 }
1973 }
1974
1975 /* Output a literal byte */
1976 s->match_length = 0;
1977 Tracevv((stderr,"%c", s->window[s->strstart]));
1978 _tr_tally_lit (s, s->window[s->strstart], bflush);
1979 s->lookahead--;
1980 s->strstart++;
1981 if (bflush) FLUSH_BLOCK(s, 0);
1982 }
1983 s->insert = 0;
1984 if (flush == Z_FINISH) {
1985 FLUSH_BLOCK(s, 1);
1986 return finish_done;
1987 }
1988 if (s->last_lit)
1989 FLUSH_BLOCK(s, 0);
1990 return block_done;
1991 }