1/* deflate.c -- compress data using the deflation algorithm 2 * Copyright (C) 1995-2024 Jean-loup Gailly and Mark Adler 3 * For conditions of distribution and use, see copyright notice in zlib.h 4 */ 5 6/* 7 * ALGORITHM 8 * 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). 12 * 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. 30 * 31 * ACKNOWLEDGEMENTS 32 * 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. 36 * 37 * REFERENCES 38 * 39 * Deutsch, L.P.,"DEFLATE Compressed Data Format Specification". 40 * Available in http://tools.ietf.org/html/rfc1951 41 * 42 * A description of the Rabin and Karp algorithm is given in the book 43 * "Algorithms" by R. Sedgewick, Addison-Wesley, p252. 44 * 45 * Fiala,E.R., and Greene,D.H. 46 * Data Compression with Finite Windows, Comm.ACM, 32,4 (1989) 490-595 47 * 48 */ 49 50#include "deflate.h" 51 52/* 53 If you use the zlib library in a product, an acknowledgment is welcome 54 in the documentation of your product. If for some reason you cannot 55 include such an acknowledgment, I would appreciate that you keep this 56 copyright string in the executable of your product. 57 */ 58 59typedef enum { 60 need_more, /* block not completed, need more input or more output */ 61 block_done, /* block flush performed */ 62 finish_started, /* finish started, need only more output at next deflate */ 63 finish_done /* finish done, accept no more input or output */ 64} block_state; 65 66typedef block_state (*compress_func)(deflate_state *s, int flush); 67/* Compression function. Returns the block state after the call. */ 68 69local block_state deflate_stored(deflate_state *s, int flush); 70local block_state deflate_fast(deflate_state *s, int flush); 71#ifndef FASTEST 72local block_state deflate_slow(deflate_state *s, int flush); 73#endif 74local block_state deflate_rle(deflate_state *s, int flush); 75local block_state deflate_huff(deflate_state *s, int flush); 76 77/* =========================================================================== 78 * Local data 79 */ 80 81#define NIL 0 82/* Tail of hash chains */ 83 84#ifndef TOO_FAR 85# define TOO_FAR 4096 86#endif 87/* Matches of length 3 are discarded if their distance exceeds TOO_FAR */ 88 89/* Values for max_lazy_match, good_match and max_chain_length, depending on 90 * the desired pack level (0..9). The values given below have been tuned to 91 * exclude worst case performance for pathological files. Better values may be 92 * found for specific files. 93 */ 94typedef struct config_s { 95 ush good_length; /* reduce lazy search above this match length */ 96 ush max_lazy; /* do not perform lazy search above this match length */ 97 ush nice_length; /* quit search above this match length */ 98 ush max_chain; 99 compress_func func; 100} config; 101 102#ifdef FASTEST 103local const config configuration_table[2] = { 104/* good lazy nice chain */ 105/* 0 */ {0, 0, 0, 0, deflate_stored}, /* store only */ 106/* 1 */ {4, 4, 8, 4, deflate_fast}}; /* max speed, no lazy matches */ 107#else 108local const config configuration_table[10] = { 109/* good lazy nice chain */ 110/* 0 */ {0, 0, 0, 0, deflate_stored}, /* store only */ 111/* 1 */ {4, 4, 8, 4, deflate_fast}, /* max speed, no lazy matches */ 112/* 2 */ {4, 5, 16, 8, deflate_fast}, 113/* 3 */ {4, 6, 32, 32, deflate_fast}, 114 115/* 4 */ {4, 4, 16, 16, deflate_slow}, /* lazy matches */ 116/* 5 */ {8, 16, 32, 32, deflate_slow}, 117/* 6 */ {8, 16, 128, 128, deflate_slow}, 118/* 7 */ {8, 32, 128, 256, deflate_slow}, 119/* 8 */ {32, 128, 258, 1024, deflate_slow}, 120/* 9 */ {32, 258, 258, 4096, deflate_slow}}; /* max compression */ 121#endif 122 123/* Note: the deflate() code requires max_lazy >= MIN_MATCH and max_chain >= 4 124 * For deflate_fast() (levels <= 3) good is ignored and lazy has a different 125 * meaning. 126 */ 127 128/* rank Z_BLOCK between Z_NO_FLUSH and Z_PARTIAL_FLUSH */ 129#define RANK(f) (((f) * 2) - ((f) > 4 ? 9 : 0)) 130 131/* =========================================================================== 132 * Update a hash value with the given input byte 133 * IN assertion: all calls to UPDATE_HASH are made with consecutive input 134 * characters, so that a running hash key can be computed from the previous 135 * key instead of complete recalculation each time. 136 */ 137#define UPDATE_HASH(s,h,c) (h = (((h) << s->hash_shift) ^ (c)) & s->hash_mask) 138 139 140/* =========================================================================== 141 * Insert string str in the dictionary and set match_head to the previous head 142 * of the hash chain (the most recent string with same hash key). Return 143 * the previous length of the hash chain. 144 * If this file is compiled with -DFASTEST, the compression level is forced 145 * to 1, and no hash chains are maintained. 146 * IN assertion: all calls to INSERT_STRING are made with consecutive input 147 * characters and the first MIN_MATCH bytes of str are valid (except for 148 * the last MIN_MATCH-1 bytes of the input file). 149 */ 150#ifdef FASTEST 151#define INSERT_STRING(s, str, match_head) \ 152 (UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \ 153 match_head = s->head[s->ins_h], \ 154 s->head[s->ins_h] = (Pos)(str)) 155#else 156#define INSERT_STRING(s, str, match_head) \ 157 (UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \ 158 match_head = s->prev[(str) & s->w_mask] = s->head[s->ins_h], \ 159 s->head[s->ins_h] = (Pos)(str)) 160#endif 161 162/* =========================================================================== 163 * Initialize the hash table (avoiding 64K overflow for 16 bit systems). 164 * prev[] will be initialized on the fly. 165 */ 166#define CLEAR_HASH(s) \ 167 do { \ 168 s->head[s->hash_size - 1] = NIL; \ 169 zmemzero((Bytef *)s->head, \ 170 (unsigned)(s->hash_size - 1)*sizeof(*s->head)); \ 171 } while (0) 172 173/* =========================================================================== 174 * Slide the hash table when sliding the window down (could be avoided with 32 175 * bit values at the expense of memory usage). We slide even when level == 0 to 176 * keep the hash table consistent if we switch back to level > 0 later. 177 */ 178#if defined(__has_feature) 179# if __has_feature(memory_sanitizer) 180 __attribute__((no_sanitize("memory"))) 181# endif 182#endif 183local void slide_hash(deflate_state *s) { 184 unsigned n, m; 185 Posf *p; 186 uInt wsize = s->w_size; 187 188 n = s->hash_size; 189 p = &s->head[n]; 190 do { 191 m = *--p; 192 *p = (Pos)(m >= wsize ? m - wsize : NIL); 193 } while (--n); 194 n = wsize; 195#ifndef FASTEST 196 p = &s->prev[n]; 197 do { 198 m = *--p; 199 *p = (Pos)(m >= wsize ? m - wsize : NIL); 200 /* If n is not on any hash chain, prev[n] is garbage but 201 * its value will never be used. 202 */ 203 } while (--n); 204#endif 205} 206 207/* =========================================================================== 208 * Read a new buffer from the current input stream, update the adler32 209 * and total number of bytes read. All deflate() input goes through 210 * this function so some applications may wish to modify it to avoid 211 * allocating a large strm->next_in buffer and copying from it. 212 * (See also flush_pending()). 213 */ 214local unsigned read_buf(z_streamp strm, Bytef *buf, unsigned size) { 215 unsigned len = strm->avail_in; 216 217 if (len > size) len = size; 218 if (len == 0) return 0; 219 220 strm->avail_in -= len; 221 222 zmemcpy(buf, strm->next_in, len); 223 if (strm->state->wrap == 1) { 224 strm->adler = adler32(strm->adler, buf, len); 225 } 226#ifdef GZIP 227 else if (strm->state->wrap == 2) { 228 strm->adler = crc32(strm->adler, buf, len); 229 } 230#endif 231 strm->next_in += len; 232 strm->total_in += len; 233 234 return len; 235} 236 237/* =========================================================================== 238 * Fill the window when the lookahead becomes insufficient. 239 * Updates strstart and lookahead. 240 * 241 * IN assertion: lookahead < MIN_LOOKAHEAD 242 * OUT assertions: strstart <= window_size-MIN_LOOKAHEAD 243 * At least one byte has been read, or avail_in == 0; reads are 244 * performed for at least two bytes (required for the zip translate_eol 245 * option -- not supported here). 246 */ 247local void fill_window(deflate_state *s) { 248 unsigned n; 249 unsigned more; /* Amount of free space at the end of the window. */ 250 uInt wsize = s->w_size; 251 252 Assert(s->lookahead < MIN_LOOKAHEAD, "already enough lookahead"); 253 254 do { 255 more = (unsigned)(s->window_size -(ulg)s->lookahead -(ulg)s->strstart); 256 257 /* Deal with !@#$% 64K limit: */ 258 if (sizeof(int) <= 2) { 259 if (more == 0 && s->strstart == 0 && s->lookahead == 0) { 260 more = wsize; 261 262 } else if (more == (unsigned)(-1)) { 263 /* Very unlikely, but possible on 16 bit machine if 264 * strstart == 0 && lookahead == 1 (input done a byte at time) 265 */ 266 more--; 267 } 268 } 269 270 /* If the window is almost full and there is insufficient lookahead, 271 * move the upper half to the lower one to make room in the upper half. 272 */ 273 if (s->strstart >= wsize + MAX_DIST(s)) { 274 275 zmemcpy(s->window, s->window + wsize, (unsigned)wsize - more); 276 s->match_start -= wsize; 277 s->strstart -= wsize; /* we now have strstart >= MAX_DIST */ 278 s->block_start -= (long) wsize; 279 if (s->insert > s->strstart) 280 s->insert = s->strstart; 281 slide_hash(s); 282 more += wsize; 283 } 284 if (s->strm->avail_in == 0) break; 285 286 /* If there was no sliding: 287 * strstart <= WSIZE+MAX_DIST-1 && lookahead <= MIN_LOOKAHEAD - 1 && 288 * more == window_size - lookahead - strstart 289 * => more >= window_size - (MIN_LOOKAHEAD-1 + WSIZE + MAX_DIST-1) 290 * => more >= window_size - 2*WSIZE + 2 291 * In the BIG_MEM or MMAP case (not yet supported), 292 * window_size == input_size + MIN_LOOKAHEAD && 293 * strstart + s->lookahead <= input_size => more >= MIN_LOOKAHEAD. 294 * Otherwise, window_size == 2*WSIZE so more >= 2. 295 * If there was sliding, more >= WSIZE. So in all cases, more >= 2. 296 */ 297 Assert(more >= 2, "more < 2"); 298 299 n = read_buf(s->strm, s->window + s->strstart + s->lookahead, more); 300 s->lookahead += n; 301 302 /* Initialize the hash value now that we have some input: */ 303 if (s->lookahead + s->insert >= MIN_MATCH) { 304 uInt str = s->strstart - s->insert; 305 s->ins_h = s->window[str]; 306 UPDATE_HASH(s, s->ins_h, s->window[str + 1]); 307#if MIN_MATCH != 3 308 Call UPDATE_HASH() MIN_MATCH-3 more times 309#endif 310 while (s->insert) { 311 UPDATE_HASH(s, s->ins_h, s->window[str + MIN_MATCH-1]); 312#ifndef FASTEST 313 s->prev[str & s->w_mask] = s->head[s->ins_h]; 314#endif 315 s->head[s->ins_h] = (Pos)str; 316 str++; 317 s->insert--; 318 if (s->lookahead + s->insert < MIN_MATCH) 319 break; 320 } 321 } 322 /* If the whole input has less than MIN_MATCH bytes, ins_h is garbage, 323 * but this is not important since only literal bytes will be emitted. 324 */ 325 326 } while (s->lookahead < MIN_LOOKAHEAD && s->strm->avail_in != 0); 327 328 /* If the WIN_INIT bytes after the end of the current data have never been 329 * written, then zero those bytes in order to avoid memory check reports of 330 * the use of uninitialized (or uninitialised as Julian writes) bytes by 331 * the longest match routines. Update the high water mark for the next 332 * time through here. WIN_INIT is set to MAX_MATCH since the longest match 333 * routines allow scanning to strstart + MAX_MATCH, ignoring lookahead. 334 */ 335 if (s->high_water < s->window_size) { 336 ulg curr = s->strstart + (ulg)(s->lookahead); 337 ulg init; 338 339 if (s->high_water < curr) { 340 /* Previous high water mark below current data -- zero WIN_INIT 341 * bytes or up to end of window, whichever is less. 342 */ 343 init = s->window_size - curr; 344 if (init > WIN_INIT) 345 init = WIN_INIT; 346 zmemzero(s->window + curr, (unsigned)init); 347 s->high_water = curr + init; 348 } 349 else if (s->high_water < (ulg)curr + WIN_INIT) { 350 /* High water mark at or above current data, but below current data 351 * plus WIN_INIT -- zero out to current data plus WIN_INIT, or up 352 * to end of window, whichever is less. 353 */ 354 init = (ulg)curr + WIN_INIT - s->high_water; 355 if (init > s->window_size - s->high_water) 356 init = s->window_size - s->high_water; 357 zmemzero(s->window + s->high_water, (unsigned)init); 358 s->high_water += init; 359 } 360 } 361 362 Assert((ulg)s->strstart <= s->window_size - MIN_LOOKAHEAD, 363 "not enough room for search"); 364} 365 366/* ========================================================================= */ 367int ZEXPORT deflateInit_(z_streamp strm, int level, const char *version, 368 int stream_size) { 369 return deflateInit2_(strm, level, Z_DEFLATED, MAX_WBITS, DEF_MEM_LEVEL, 370 Z_DEFAULT_STRATEGY, version, stream_size); 371 /* To do: ignore strm->next_in if we use it as window */ 372} 373 374/* ========================================================================= */ 375int ZEXPORT deflateInit2_(z_streamp strm, int level, int method, 376 int windowBits, int memLevel, int strategy, 377 const char *version, int stream_size) { 378 deflate_state *s; 379 int wrap = 1; 380 static const char my_version[] = ZLIB_VERSION; 381 382 if (version == Z_NULL || version[0] != my_version[0] || 383 stream_size != sizeof(z_stream)) { 384 return Z_VERSION_ERROR; 385 } 386 if (strm == Z_NULL) return Z_STREAM_ERROR; 387 388 strm->msg = Z_NULL; 389 if (strm->zalloc == (alloc_func)0) { 390#ifdef Z_SOLO 391 return Z_STREAM_ERROR; 392#else 393 strm->zalloc = zcalloc; 394 strm->opaque = (voidpf)0; 395#endif 396 } 397 if (strm->zfree == (free_func)0) 398#ifdef Z_SOLO 399 return Z_STREAM_ERROR; 400#else 401 strm->zfree = zcfree; 402#endif 403 404#ifdef FASTEST 405 if (level != 0) level = 1; 406#else 407 if (level == Z_DEFAULT_COMPRESSION) level = 6; 408#endif 409 410 if (windowBits < 0) { /* suppress zlib wrapper */ 411 wrap = 0; 412 if (windowBits < -15) 413 return Z_STREAM_ERROR; 414 windowBits = -windowBits; 415 } 416#ifdef GZIP 417 else if (windowBits > 15) { 418 wrap = 2; /* write gzip wrapper instead */ 419 windowBits -= 16; 420 } 421#endif 422 if (memLevel < 1 || memLevel > MAX_MEM_LEVEL || method != Z_DEFLATED || 423 windowBits < 8 || windowBits > 15 || level < 0 || level > 9 || 424 strategy < 0 || strategy > Z_FIXED || (windowBits == 8 && wrap != 1)) { 425 return Z_STREAM_ERROR; 426 } 427 if (windowBits == 8) windowBits = 9; /* until 256-byte window bug fixed */ 428 s = (deflate_state *) ZALLOC(strm, 1, sizeof(deflate_state)); 429 if (s == Z_NULL) return Z_MEM_ERROR; 430 strm->state = (struct internal_state FAR *)s; 431 s->strm = strm; 432 s->status = INIT_STATE; /* to pass state test in deflateReset() */ 433 434 s->wrap = wrap; 435 s->gzhead = Z_NULL; 436 s->w_bits = (uInt)windowBits; 437 s->w_size = 1 << s->w_bits; 438 s->w_mask = s->w_size - 1; 439 440 s->hash_bits = (uInt)memLevel + 7; 441 s->hash_size = 1 << s->hash_bits; 442 s->hash_mask = s->hash_size - 1; 443 s->hash_shift = ((s->hash_bits + MIN_MATCH-1) / MIN_MATCH); 444 445 s->window = (Bytef *) ZALLOC(strm, s->w_size, 2*sizeof(Byte)); 446 s->prev = (Posf *) ZALLOC(strm, s->w_size, sizeof(Pos)); 447 s->head = (Posf *) ZALLOC(strm, s->hash_size, sizeof(Pos)); 448 449 s->high_water = 0; /* nothing written to s->window yet */ 450 451 s->lit_bufsize = 1 << (memLevel + 6); /* 16K elements by default */ 452 453 /* We overlay pending_buf and sym_buf. This works since the average size 454 * for length/distance pairs over any compressed block is assured to be 31 455 * bits or less. 456 * 457 * Analysis: The longest fixed codes are a length code of 8 bits plus 5 458 * extra bits, for lengths 131 to 257. The longest fixed distance codes are 459 * 5 bits plus 13 extra bits, for distances 16385 to 32768. The longest 460 * possible fixed-codes length/distance pair is then 31 bits total. 461 * 462 * sym_buf starts one-fourth of the way into pending_buf. So there are 463 * three bytes in sym_buf for every four bytes in pending_buf. Each symbol 464 * in sym_buf is three bytes -- two for the distance and one for the 465 * literal/length. As each symbol is consumed, the pointer to the next 466 * sym_buf value to read moves forward three bytes. From that symbol, up to 467 * 31 bits are written to pending_buf. The closest the written pending_buf 468 * bits gets to the next sym_buf symbol to read is just before the last 469 * code is written. At that time, 31*(n - 2) bits have been written, just 470 * after 24*(n - 2) bits have been consumed from sym_buf. sym_buf starts at 471 * 8*n bits into pending_buf. (Note that the symbol buffer fills when n - 1 472 * symbols are written.) The closest the writing gets to what is unread is 473 * then n + 14 bits. Here n is lit_bufsize, which is 16384 by default, and 474 * can range from 128 to 32768. 475 * 476 * Therefore, at a minimum, there are 142 bits of space between what is 477 * written and what is read in the overlain buffers, so the symbols cannot 478 * be overwritten by the compressed data. That space is actually 139 bits, 479 * due to the three-bit fixed-code block header. 480 * 481 * That covers the case where either Z_FIXED is specified, forcing fixed 482 * codes, or when the use of fixed codes is chosen, because that choice 483 * results in a smaller compressed block than dynamic codes. That latter 484 * condition then assures that the above analysis also covers all dynamic 485 * blocks. A dynamic-code block will only be chosen to be emitted if it has 486 * fewer bits than a fixed-code block would for the same set of symbols. 487 * Therefore its average symbol length is assured to be less than 31. So 488 * the compressed data for a dynamic block also cannot overwrite the 489 * symbols from which it is being constructed. 490 */ 491 492 s->pending_buf = (uchf *) ZALLOC(strm, s->lit_bufsize, LIT_BUFS); 493 s->pending_buf_size = (ulg)s->lit_bufsize * 4; 494 495 if (s->window == Z_NULL || s->prev == Z_NULL || s->head == Z_NULL || 496 s->pending_buf == Z_NULL) { 497 s->status = FINISH_STATE; 498 strm->msg = ERR_MSG(Z_MEM_ERROR); 499 deflateEnd (strm); 500 return Z_MEM_ERROR; 501 } 502#ifdef LIT_MEM 503 s->d_buf = (ushf *)(s->pending_buf + (s->lit_bufsize << 1)); 504 s->l_buf = s->pending_buf + (s->lit_bufsize << 2); 505 s->sym_end = s->lit_bufsize - 1; 506#else 507 s->sym_buf = s->pending_buf + s->lit_bufsize; 508 s->sym_end = (s->lit_bufsize - 1) * 3; 509#endif 510 /* We avoid equality with lit_bufsize*3 because of wraparound at 64K 511 * on 16 bit machines and because stored blocks are restricted to 512 * 64K-1 bytes. 513 */ 514 515 s->level = level; 516 s->strategy = strategy; 517 s->method = (Byte)method; 518 519 return deflateReset(strm); 520} 521 522/* ========================================================================= 523 * Check for a valid deflate stream state. Return 0 if ok, 1 if not. 524 */ 525local int deflateStateCheck(z_streamp strm) { 526 deflate_state *s; 527 if (strm == Z_NULL || 528 strm->zalloc == (alloc_func)0 || strm->zfree == (free_func)0) 529 return 1; 530 s = strm->state; 531 if (s == Z_NULL || s->strm != strm || (s->status != INIT_STATE && 532#ifdef GZIP 533 s->status != GZIP_STATE && 534#endif 535 s->status != EXTRA_STATE && 536 s->status != NAME_STATE && 537 s->status != COMMENT_STATE && 538 s->status != HCRC_STATE && 539 s->status != BUSY_STATE && 540 s->status != FINISH_STATE)) 541 return 1; 542 return 0; 543} 544 545/* ========================================================================= */ 546int ZEXPORT deflateSetDictionary(z_streamp strm, const Bytef *dictionary, 547 uInt dictLength) { 548 deflate_state *s; 549 uInt str, n; 550 int wrap; 551 unsigned avail; 552 z_const unsigned char *next; 553 554 if (deflateStateCheck(strm) || dictionary == Z_NULL) 555 return Z_STREAM_ERROR; 556 s = strm->state; 557 wrap = s->wrap; 558 if (wrap == 2 || (wrap == 1 && s->status != INIT_STATE) || s->lookahead) 559 return Z_STREAM_ERROR; 560 561 /* when using zlib wrappers, compute Adler-32 for provided dictionary */ 562 if (wrap == 1) 563 strm->adler = adler32(strm->adler, dictionary, dictLength); 564 s->wrap = 0; /* avoid computing Adler-32 in read_buf */ 565 566 /* if dictionary would fill window, just replace the history */ 567 if (dictLength >= s->w_size) { 568 if (wrap == 0) { /* already empty otherwise */ 569 CLEAR_HASH(s); 570 s->strstart = 0; 571 s->block_start = 0L; 572 s->insert = 0; 573 } 574 dictionary += dictLength - s->w_size; /* use the tail */ 575 dictLength = s->w_size; 576 } 577 578 /* insert dictionary into window and hash */ 579 avail = strm->avail_in; 580 next = strm->next_in; 581 strm->avail_in = dictLength; 582 strm->next_in = (z_const Bytef *)dictionary; 583 fill_window(s); 584 while (s->lookahead >= MIN_MATCH) { 585 str = s->strstart; 586 n = s->lookahead - (MIN_MATCH-1); 587 do { 588 UPDATE_HASH(s, s->ins_h, s->window[str + MIN_MATCH-1]); 589#ifndef FASTEST 590 s->prev[str & s->w_mask] = s->head[s->ins_h]; 591#endif 592 s->head[s->ins_h] = (Pos)str; 593 str++; 594 } while (--n); 595 s->strstart = str; 596 s->lookahead = MIN_MATCH-1; 597 fill_window(s); 598 } 599 s->strstart += s->lookahead; 600 s->block_start = (long)s->strstart; 601 s->insert = s->lookahead; 602 s->lookahead = 0; 603 s->match_length = s->prev_length = MIN_MATCH-1; 604 s->match_available = 0; 605 strm->next_in = next; 606 strm->avail_in = avail; 607 s->wrap = wrap; 608 return Z_OK; 609} 610 611/* ========================================================================= */ 612int ZEXPORT deflateGetDictionary(z_streamp strm, Bytef *dictionary, 613 uInt *dictLength) { 614 deflate_state *s; 615 uInt len; 616 617 if (deflateStateCheck(strm)) 618 return Z_STREAM_ERROR; 619 s = strm->state; 620 len = s->strstart + s->lookahead; 621 if (len > s->w_size) 622 len = s->w_size; 623 if (dictionary != Z_NULL && len) 624 zmemcpy(dictionary, s->window + s->strstart + s->lookahead - len, len); 625 if (dictLength != Z_NULL) 626 *dictLength = len; 627 return Z_OK; 628} 629 630/* ========================================================================= */ 631int ZEXPORT deflateResetKeep(z_streamp strm) { 632 deflate_state *s; 633 634 if (deflateStateCheck(strm)) { 635 return Z_STREAM_ERROR; 636 } 637 638 strm->total_in = strm->total_out = 0; 639 strm->msg = Z_NULL; /* use zfree if we ever allocate msg dynamically */ 640 strm->data_type = Z_UNKNOWN; 641 642 s = (deflate_state *)strm->state; 643 s->pending = 0; 644 s->pending_out = s->pending_buf; 645 646 if (s->wrap < 0) { 647 s->wrap = -s->wrap; /* was made negative by deflate(..., Z_FINISH); */ 648 } 649 s->status = 650#ifdef GZIP 651 s->wrap == 2 ? GZIP_STATE : 652#endif 653 INIT_STATE; 654 strm->adler = 655#ifdef GZIP 656 s->wrap == 2 ? crc32(0L, Z_NULL, 0) : 657#endif 658 adler32(0L, Z_NULL, 0); 659 s->last_flush = -2; 660 661 _tr_init(s); 662 663 return Z_OK; 664} 665 666/* =========================================================================== 667 * Initialize the "longest match" routines for a new zlib stream 668 */ 669local void lm_init(deflate_state *s) { 670 s->window_size = (ulg)2L*s->w_size; 671 672 CLEAR_HASH(s); 673 674 /* Set the default configuration parameters: 675 */ 676 s->max_lazy_match = configuration_table[s->level].max_lazy; 677 s->good_match = configuration_table[s->level].good_length; 678 s->nice_match = configuration_table[s->level].nice_length; 679 s->max_chain_length = configuration_table[s->level].max_chain; 680 681 s->strstart = 0; 682 s->block_start = 0L; 683 s->lookahead = 0; 684 s->insert = 0; 685 s->match_length = s->prev_length = MIN_MATCH-1; 686 s->match_available = 0; 687 s->ins_h = 0; 688} 689 690/* ========================================================================= */ 691int ZEXPORT deflateReset(z_streamp strm) { 692 int ret; 693 694 ret = deflateResetKeep(strm); 695 if (ret == Z_OK) 696 lm_init(strm->state); 697 return ret; 698} 699 700/* ========================================================================= */ 701int ZEXPORT deflateSetHeader(z_streamp strm, gz_headerp head) { 702 if (deflateStateCheck(strm) || strm->state->wrap != 2) 703 return Z_STREAM_ERROR; 704 strm->state->gzhead = head; 705 return Z_OK; 706} 707 708/* ========================================================================= */ 709int ZEXPORT deflatePending(z_streamp strm, unsigned *pending, int *bits) { 710 if (deflateStateCheck(strm)) return Z_STREAM_ERROR; 711 if (pending != Z_NULL) 712 *pending = strm->state->pending; 713 if (bits != Z_NULL) 714 *bits = strm->state->bi_valid; 715 return Z_OK; 716} 717 718/* ========================================================================= */ 719int ZEXPORT deflatePrime(z_streamp strm, int bits, int value) { 720 deflate_state *s; 721 int put; 722 723 if (deflateStateCheck(strm)) return Z_STREAM_ERROR; 724 s = strm->state; 725#ifdef LIT_MEM 726 if (bits < 0 || bits > 16 || 727 (uchf *)s->d_buf < s->pending_out + ((Buf_size + 7) >> 3)) 728 return Z_BUF_ERROR; 729#else 730 if (bits < 0 || bits > 16 || 731 s->sym_buf < s->pending_out + ((Buf_size + 7) >> 3)) 732 return Z_BUF_ERROR; 733#endif 734 do { 735 put = Buf_size - s->bi_valid; 736 if (put > bits) 737 put = bits; 738 s->bi_buf |= (ush)((value & ((1 << put) - 1)) << s->bi_valid); 739 s->bi_valid += put; 740 _tr_flush_bits(s); 741 value >>= put; 742 bits -= put; 743 } while (bits); 744 return Z_OK; 745} 746 747/* ========================================================================= */ 748int ZEXPORT deflateParams(z_streamp strm, int level, int strategy) { 749 deflate_state *s; 750 compress_func func; 751 752 if (deflateStateCheck(strm)) return Z_STREAM_ERROR; 753 s = strm->state; 754 755#ifdef FASTEST 756 if (level != 0) level = 1; 757#else 758 if (level == Z_DEFAULT_COMPRESSION) level = 6; 759#endif 760 if (level < 0 || level > 9 || strategy < 0 || strategy > Z_FIXED) { 761 return Z_STREAM_ERROR; 762 } 763 func = configuration_table[s->level].func; 764 765 if ((strategy != s->strategy || func != configuration_table[level].func) && 766 s->last_flush != -2) { 767 /* Flush the last buffer: */ 768 int err = deflate(strm, Z_BLOCK); 769 if (err == Z_STREAM_ERROR) 770 return err; 771 if (strm->avail_in || (s->strstart - s->block_start) + s->lookahead) 772 return Z_BUF_ERROR; 773 } 774 if (s->level != level) { 775 if (s->level == 0 && s->matches != 0) { 776 if (s->matches == 1) 777 slide_hash(s); 778 else 779 CLEAR_HASH(s); 780 s->matches = 0; 781 } 782 s->level = level; 783 s->max_lazy_match = configuration_table[level].max_lazy; 784 s->good_match = configuration_table[level].good_length; 785 s->nice_match = configuration_table[level].nice_length; 786 s->max_chain_length = configuration_table[level].max_chain; 787 } 788 s->strategy = strategy; 789 return Z_OK; 790} 791 792/* ========================================================================= */ 793int ZEXPORT deflateTune(z_streamp strm, int good_length, int max_lazy, 794 int nice_length, int max_chain) { 795 deflate_state *s; 796 797 if (deflateStateCheck(strm)) return Z_STREAM_ERROR; 798 s = strm->state; 799 s->good_match = (uInt)good_length; 800 s->max_lazy_match = (uInt)max_lazy; 801 s->nice_match = nice_length; 802 s->max_chain_length = (uInt)max_chain; 803 return Z_OK; 804} 805 806/* ========================================================================= 807 * For the default windowBits of 15 and memLevel of 8, this function returns a 808 * close to exact, as well as small, upper bound on the compressed size. This 809 * is an expansion of ~0.03%, plus a small constant. 810 * 811 * For any setting other than those defaults for windowBits and memLevel, one 812 * of two worst case bounds is returned. This is at most an expansion of ~4% or 813 * ~13%, plus a small constant. 814 * 815 * Both the 0.03% and 4% derive from the overhead of stored blocks. The first 816 * one is for stored blocks of 16383 bytes (memLevel == 8), whereas the second 817 * is for stored blocks of 127 bytes (the worst case memLevel == 1). The 818 * expansion results from five bytes of header for each stored block. 819 * 820 * The larger expansion of 13% results from a window size less than or equal to 821 * the symbols buffer size (windowBits <= memLevel + 7). In that case some of 822 * the data being compressed may have slid out of the sliding window, impeding 823 * a stored block from being emitted. Then the only choice is a fixed or 824 * dynamic block, where a fixed block limits the maximum expansion to 9 bits 825 * per 8-bit byte, plus 10 bits for every block. The smallest block size for 826 * which this can occur is 255 (memLevel == 2). 827 * 828 * Shifts are used to approximate divisions, for speed. 829 */ 830uLong ZEXPORT deflateBound(z_streamp strm, uLong sourceLen) { 831 deflate_state *s; 832 uLong fixedlen, storelen, wraplen; 833 834 /* upper bound for fixed blocks with 9-bit literals and length 255 835 (memLevel == 2, which is the lowest that may not use stored blocks) -- 836 ~13% overhead plus a small constant */ 837 fixedlen = sourceLen + (sourceLen >> 3) + (sourceLen >> 8) + 838 (sourceLen >> 9) + 4; 839 840 /* upper bound for stored blocks with length 127 (memLevel == 1) -- 841 ~4% overhead plus a small constant */ 842 storelen = sourceLen + (sourceLen >> 5) + (sourceLen >> 7) + 843 (sourceLen >> 11) + 7; 844 845 /* if can't get parameters, return larger bound plus a wrapper */ 846 if (deflateStateCheck(strm)) 847 return (fixedlen > storelen ? fixedlen : storelen) + 18; 848 849 /* compute wrapper length */ 850 s = strm->state; 851 switch (s->wrap < 0 ? -s->wrap : s->wrap) { 852 case 0: /* raw deflate */ 853 wraplen = 0; 854 break; 855 case 1: /* zlib wrapper */ 856 wraplen = 6 + (s->strstart ? 4 : 0); 857 break; 858#ifdef GZIP 859 case 2: /* gzip wrapper */ 860 wraplen = 18; 861 if (s->gzhead != Z_NULL) { /* user-supplied gzip header */ 862 Bytef *str; 863 if (s->gzhead->extra != Z_NULL) 864 wraplen += 2 + s->gzhead->extra_len; 865 str = s->gzhead->name; 866 if (str != Z_NULL) 867 do { 868 wraplen++; 869 } while (*str++); 870 str = s->gzhead->comment; 871 if (str != Z_NULL) 872 do { 873 wraplen++; 874 } while (*str++); 875 if (s->gzhead->hcrc) 876 wraplen += 2; 877 } 878 break; 879#endif 880 default: /* for compiler happiness */ 881 wraplen = 18; 882 } 883 884 /* if not default parameters, return one of the conservative bounds */ 885 if (s->w_bits != 15 || s->hash_bits != 8 + 7) 886 return (s->w_bits <= s->hash_bits && s->level ? fixedlen : storelen) + 887 wraplen; 888 889 /* default settings: return tight bound for that case -- ~0.03% overhead 890 plus a small constant */ 891 return sourceLen + (sourceLen >> 12) + (sourceLen >> 14) + 892 (sourceLen >> 25) + 13 - 6 + wraplen; 893} 894 895/* ========================================================================= 896 * Put a short in the pending buffer. The 16-bit value is put in MSB order. 897 * IN assertion: the stream state is correct and there is enough room in 898 * pending_buf. 899 */ 900local void putShortMSB(deflate_state *s, uInt b) { 901 put_byte(s, (Byte)(b >> 8)); 902 put_byte(s, (Byte)(b & 0xff)); 903} 904 905/* ========================================================================= 906 * Flush as much pending output as possible. All deflate() output, except for 907 * some deflate_stored() output, goes through this function so some 908 * applications may wish to modify it to avoid allocating a large 909 * strm->next_out buffer and copying into it. (See also read_buf()). 910 */ 911local void flush_pending(z_streamp strm) { 912 unsigned len; 913 deflate_state *s = strm->state; 914 915 _tr_flush_bits(s); 916 len = s->pending; 917 if (len > strm->avail_out) len = strm->avail_out; 918 if (len == 0) return; 919 920 zmemcpy(strm->next_out, s->pending_out, len); 921 strm->next_out += len; 922 s->pending_out += len; 923 strm->total_out += len; 924 strm->avail_out -= len; 925 s->pending -= len; 926 if (s->pending == 0) { 927 s->pending_out = s->pending_buf; 928 } 929} 930 931/* =========================================================================== 932 * Update the header CRC with the bytes s->pending_buf[beg..s->pending - 1]. 933 */ 934#define HCRC_UPDATE(beg) \ 935 do { \ 936 if (s->gzhead->hcrc && s->pending > (beg)) \ 937 strm->adler = crc32(strm->adler, s->pending_buf + (beg), \ 938 s->pending - (beg)); \ 939 } while (0) 940 941/* ========================================================================= */ 942int ZEXPORT deflate(z_streamp strm, int flush) { 943 int old_flush; /* value of flush param for previous deflate call */ 944 deflate_state *s; 945 946 if (deflateStateCheck(strm) || flush > Z_BLOCK || flush < 0) { 947 return Z_STREAM_ERROR; 948 } 949 s = strm->state; 950 951 if (strm->next_out == Z_NULL || 952 (strm->avail_in != 0 && strm->next_in == Z_NULL) || 953 (s->status == FINISH_STATE && flush != Z_FINISH)) { 954 ERR_RETURN(strm, Z_STREAM_ERROR); 955 } 956 if (strm->avail_out == 0) ERR_RETURN(strm, Z_BUF_ERROR); 957 958 old_flush = s->last_flush; 959 s->last_flush = flush; 960 961 /* Flush as much pending output as possible */ 962 if (s->pending != 0) { 963 flush_pending(strm); 964 if (strm->avail_out == 0) { 965 /* Since avail_out is 0, deflate will be called again with 966 * more output space, but possibly with both pending and 967 * avail_in equal to zero. There won't be anything to do, 968 * but this is not an error situation so make sure we 969 * return OK instead of BUF_ERROR at next call of deflate: 970 */ 971 s->last_flush = -1; 972 return Z_OK; 973 } 974 975 /* Make sure there is something to do and avoid duplicate consecutive 976 * flushes. For repeated and useless calls with Z_FINISH, we keep 977 * returning Z_STREAM_END instead of Z_BUF_ERROR. 978 */ 979 } else if (strm->avail_in == 0 && RANK(flush) <= RANK(old_flush) && 980 flush != Z_FINISH) { 981 ERR_RETURN(strm, Z_BUF_ERROR); 982 } 983 984 /* User must not provide more input after the first FINISH: */ 985 if (s->status == FINISH_STATE && strm->avail_in != 0) { 986 ERR_RETURN(strm, Z_BUF_ERROR); 987 } 988 989 /* Write the header */ 990 if (s->status == INIT_STATE && s->wrap == 0) 991 s->status = BUSY_STATE; 992 if (s->status == INIT_STATE) { 993 /* zlib header */ 994 uInt header = (Z_DEFLATED + ((s->w_bits - 8) << 4)) << 8; 995 uInt level_flags; 996 997 if (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2) 998 level_flags = 0; 999 else if (s->level < 6) 1000 level_flags = 1; 1001 else if (s->level == 6) 1002 level_flags = 2; 1003 else 1004 level_flags = 3; 1005 header |= (level_flags << 6); 1006 if (s->strstart != 0) header |= PRESET_DICT; 1007 header += 31 - (header % 31); 1008 1009 putShortMSB(s, header); 1010 1011 /* Save the adler32 of the preset dictionary: */ 1012 if (s->strstart != 0) { 1013 putShortMSB(s, (uInt)(strm->adler >> 16)); 1014 putShortMSB(s, (uInt)(strm->adler & 0xffff)); 1015 } 1016 strm->adler = adler32(0L, Z_NULL, 0); 1017 s->status = BUSY_STATE; 1018 1019 /* Compression must start with an empty pending buffer */ 1020 flush_pending(strm); 1021 if (s->pending != 0) { 1022 s->last_flush = -1; 1023 return Z_OK; 1024 } 1025 } 1026#ifdef GZIP 1027 if (s->status == GZIP_STATE) { 1028 /* gzip header */ 1029 strm->adler = crc32(0L, Z_NULL, 0); 1030 put_byte(s, 31); 1031 put_byte(s, 139); 1032 put_byte(s, 8); 1033 if (s->gzhead == Z_NULL) { 1034 put_byte(s, 0); 1035 put_byte(s, 0); 1036 put_byte(s, 0); 1037 put_byte(s, 0); 1038 put_byte(s, 0); 1039 put_byte(s, s->level == 9 ? 2 : 1040 (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ? 1041 4 : 0)); 1042 put_byte(s, OS_CODE); 1043 s->status = BUSY_STATE; 1044 1045 /* Compression must start with an empty pending buffer */ 1046 flush_pending(strm); 1047 if (s->pending != 0) { 1048 s->last_flush = -1; 1049 return Z_OK; 1050 } 1051 } 1052 else { 1053 put_byte(s, (s->gzhead->text ? 1 : 0) + 1054 (s->gzhead->hcrc ? 2 : 0) + 1055 (s->gzhead->extra == Z_NULL ? 0 : 4) + 1056 (s->gzhead->name == Z_NULL ? 0 : 8) + 1057 (s->gzhead->comment == Z_NULL ? 0 : 16) 1058 ); 1059 put_byte(s, (Byte)(s->gzhead->time & 0xff)); 1060 put_byte(s, (Byte)((s->gzhead->time >> 8) & 0xff)); 1061 put_byte(s, (Byte)((s->gzhead->time >> 16) & 0xff)); 1062 put_byte(s, (Byte)((s->gzhead->time >> 24) & 0xff)); 1063 put_byte(s, s->level == 9 ? 2 : 1064 (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ? 1065 4 : 0)); 1066 put_byte(s, s->gzhead->os & 0xff); 1067 if (s->gzhead->extra != Z_NULL) { 1068 put_byte(s, s->gzhead->extra_len & 0xff); 1069 put_byte(s, (s->gzhead->extra_len >> 8) & 0xff); 1070 } 1071 if (s->gzhead->hcrc) 1072 strm->adler = crc32(strm->adler, s->pending_buf, 1073 s->pending); 1074 s->gzindex = 0; 1075 s->status = EXTRA_STATE; 1076 } 1077 } 1078 if (s->status == EXTRA_STATE) { 1079 if (s->gzhead->extra != Z_NULL) { 1080 ulg beg = s->pending; /* start of bytes to update crc */ 1081 uInt left = (s->gzhead->extra_len & 0xffff) - s->gzindex; 1082 while (s->pending + left > s->pending_buf_size) { 1083 uInt copy = s->pending_buf_size - s->pending; 1084 zmemcpy(s->pending_buf + s->pending, 1085 s->gzhead->extra + s->gzindex, copy); 1086 s->pending = s->pending_buf_size; 1087 HCRC_UPDATE(beg); 1088 s->gzindex += copy; 1089 flush_pending(strm); 1090 if (s->pending != 0) { 1091 s->last_flush = -1; 1092 return Z_OK; 1093 } 1094 beg = 0; 1095 left -= copy; 1096 } 1097 zmemcpy(s->pending_buf + s->pending, 1098 s->gzhead->extra + s->gzindex, left); 1099 s->pending += left; 1100 HCRC_UPDATE(beg); 1101 s->gzindex = 0; 1102 } 1103 s->status = NAME_STATE; 1104 } 1105 if (s->status == NAME_STATE) { 1106 if (s->gzhead->name != Z_NULL) { 1107 ulg beg = s->pending; /* start of bytes to update crc */ 1108 int val; 1109 do { 1110 if (s->pending == s->pending_buf_size) { 1111 HCRC_UPDATE(beg); 1112 flush_pending(strm); 1113 if (s->pending != 0) { 1114 s->last_flush = -1; 1115 return Z_OK; 1116 } 1117 beg = 0; 1118 } 1119 val = s->gzhead->name[s->gzindex++]; 1120 put_byte(s, val); 1121 } while (val != 0); 1122 HCRC_UPDATE(beg); 1123 s->gzindex = 0; 1124 } 1125 s->status = COMMENT_STATE; 1126 } 1127 if (s->status == COMMENT_STATE) { 1128 if (s->gzhead->comment != Z_NULL) { 1129 ulg beg = s->pending; /* start of bytes to update crc */ 1130 int val; 1131 do { 1132 if (s->pending == s->pending_buf_size) { 1133 HCRC_UPDATE(beg); 1134 flush_pending(strm); 1135 if (s->pending != 0) { 1136 s->last_flush = -1; 1137 return Z_OK; 1138 } 1139 beg = 0; 1140 } 1141 val = s->gzhead->comment[s->gzindex++]; 1142 put_byte(s, val); 1143 } while (val != 0); 1144 HCRC_UPDATE(beg); 1145 } 1146 s->status = HCRC_STATE; 1147 } 1148 if (s->status == HCRC_STATE) { 1149 if (s->gzhead->hcrc) { 1150 if (s->pending + 2 > s->pending_buf_size) { 1151 flush_pending(strm); 1152 if (s->pending != 0) { 1153 s->last_flush = -1; 1154 return Z_OK; 1155 } 1156 } 1157 put_byte(s, (Byte)(strm->adler & 0xff)); 1158 put_byte(s, (Byte)((strm->adler >> 8) & 0xff)); 1159 strm->adler = crc32(0L, Z_NULL, 0); 1160 } 1161 s->status = BUSY_STATE; 1162 1163 /* Compression must start with an empty pending buffer */ 1164 flush_pending(strm); 1165 if (s->pending != 0) { 1166 s->last_flush = -1; 1167 return Z_OK; 1168 } 1169 } 1170#endif 1171 1172 /* Start a new block or continue the current one. 1173 */ 1174 if (strm->avail_in != 0 || s->lookahead != 0 || 1175 (flush != Z_NO_FLUSH && s->status != FINISH_STATE)) { 1176 block_state bstate; 1177 1178 bstate = s->level == 0 ? deflate_stored(s, flush) : 1179 s->strategy == Z_HUFFMAN_ONLY ? deflate_huff(s, flush) : 1180 s->strategy == Z_RLE ? deflate_rle(s, flush) : 1181 (*(configuration_table[s->level].func))(s, flush); 1182 1183 if (bstate == finish_started || bstate == finish_done) { 1184 s->status = FINISH_STATE; 1185 } 1186 if (bstate == need_more || bstate == finish_started) { 1187 if (strm->avail_out == 0) { 1188 s->last_flush = -1; /* avoid BUF_ERROR next call, see above */ 1189 } 1190 return Z_OK; 1191 /* If flush != Z_NO_FLUSH && avail_out == 0, the next call 1192 * of deflate should use the same flush parameter to make sure 1193 * that the flush is complete. So we don't have to output an 1194 * empty block here, this will be done at next call. This also 1195 * ensures that for a very small output buffer, we emit at most 1196 * one empty block. 1197 */ 1198 } 1199 if (bstate == block_done) { 1200 if (flush == Z_PARTIAL_FLUSH) { 1201 _tr_align(s); 1202 } else if (flush != Z_BLOCK) { /* FULL_FLUSH or SYNC_FLUSH */ 1203 _tr_stored_block(s, (char*)0, 0L, 0); 1204 /* For a full flush, this empty block will be recognized 1205 * as a special marker by inflate_sync(). 1206 */ 1207 if (flush == Z_FULL_FLUSH) { 1208 CLEAR_HASH(s); /* forget history */ 1209 if (s->lookahead == 0) { 1210 s->strstart = 0; 1211 s->block_start = 0L; 1212 s->insert = 0; 1213 } 1214 } 1215 } 1216 flush_pending(strm); 1217 if (strm->avail_out == 0) { 1218 s->last_flush = -1; /* avoid BUF_ERROR at next call, see above */ 1219 return Z_OK; 1220 } 1221 } 1222 } 1223 1224 if (flush != Z_FINISH) return Z_OK; 1225 if (s->wrap <= 0) return Z_STREAM_END; 1226 1227 /* Write the trailer */ 1228#ifdef GZIP 1229 if (s->wrap == 2) { 1230 put_byte(s, (Byte)(strm->adler & 0xff)); 1231 put_byte(s, (Byte)((strm->adler >> 8) & 0xff)); 1232 put_byte(s, (Byte)((strm->adler >> 16) & 0xff)); 1233 put_byte(s, (Byte)((strm->adler >> 24) & 0xff)); 1234 put_byte(s, (Byte)(strm->total_in & 0xff)); 1235 put_byte(s, (Byte)((strm->total_in >> 8) & 0xff)); 1236 put_byte(s, (Byte)((strm->total_in >> 16) & 0xff)); 1237 put_byte(s, (Byte)((strm->total_in >> 24) & 0xff)); 1238 } 1239 else 1240#endif 1241 { 1242 putShortMSB(s, (uInt)(strm->adler >> 16)); 1243 putShortMSB(s, (uInt)(strm->adler & 0xffff)); 1244 } 1245 flush_pending(strm); 1246 /* If avail_out is zero, the application will call deflate again 1247 * to flush the rest. 1248 */ 1249 if (s->wrap > 0) s->wrap = -s->wrap; /* write the trailer only once! */ 1250 return s->pending != 0 ? Z_OK : Z_STREAM_END; 1251} 1252 1253/* ========================================================================= */ 1254int ZEXPORT deflateEnd(z_streamp strm) { 1255 int status; 1256 1257 if (deflateStateCheck(strm)) return Z_STREAM_ERROR; 1258 1259 status = strm->state->status; 1260 1261 /* Deallocate in reverse order of allocations: */ 1262 TRY_FREE(strm, strm->state->pending_buf); 1263 TRY_FREE(strm, strm->state->head); 1264 TRY_FREE(strm, strm->state->prev); 1265 TRY_FREE(strm, strm->state->window); 1266 1267 ZFREE(strm, strm->state); 1268 strm->state = Z_NULL; 1269 1270 return status == BUSY_STATE ? Z_DATA_ERROR : Z_OK; 1271} 1272 1273/* ========================================================================= 1274 * Copy the source state to the destination state. 1275 * To simplify the source, this is not supported for 16-bit MSDOS (which 1276 * doesn't have enough memory anyway to duplicate compression states). 1277 */ 1278int ZEXPORT deflateCopy(z_streamp dest, z_streamp source) { 1279#ifdef MAXSEG_64K 1280 (void)dest; 1281 (void)source; 1282 return Z_STREAM_ERROR; 1283#else 1284 deflate_state *ds; 1285 deflate_state *ss; 1286 1287 1288 if (deflateStateCheck(source) || dest == Z_NULL) { 1289 return Z_STREAM_ERROR; 1290 } 1291 1292 ss = source->state; 1293 1294 zmemcpy((voidpf)dest, (voidpf)source, sizeof(z_stream)); 1295 1296 ds = (deflate_state *) ZALLOC(dest, 1, sizeof(deflate_state)); 1297 if (ds == Z_NULL) return Z_MEM_ERROR; 1298 dest->state = (struct internal_state FAR *) ds; 1299 zmemcpy((voidpf)ds, (voidpf)ss, sizeof(deflate_state)); 1300 ds->strm = dest; 1301 1302 ds->window = (Bytef *) ZALLOC(dest, ds->w_size, 2*sizeof(Byte)); 1303 ds->prev = (Posf *) ZALLOC(dest, ds->w_size, sizeof(Pos)); 1304 ds->head = (Posf *) ZALLOC(dest, ds->hash_size, sizeof(Pos)); 1305 ds->pending_buf = (uchf *) ZALLOC(dest, ds->lit_bufsize, LIT_BUFS); 1306 1307 if (ds->window == Z_NULL || ds->prev == Z_NULL || ds->head == Z_NULL || 1308 ds->pending_buf == Z_NULL) { 1309 deflateEnd (dest); 1310 return Z_MEM_ERROR; 1311 } 1312 /* following zmemcpy do not work for 16-bit MSDOS */ 1313 zmemcpy(ds->window, ss->window, ds->w_size * 2 * sizeof(Byte)); 1314 zmemcpy((voidpf)ds->prev, (voidpf)ss->prev, ds->w_size * sizeof(Pos)); 1315 zmemcpy((voidpf)ds->head, (voidpf)ss->head, ds->hash_size * sizeof(Pos)); 1316 zmemcpy(ds->pending_buf, ss->pending_buf, ds->lit_bufsize * LIT_BUFS); 1317 1318 ds->pending_out = ds->pending_buf + (ss->pending_out - ss->pending_buf); 1319#ifdef LIT_MEM 1320 ds->d_buf = (ushf *)(ds->pending_buf + (ds->lit_bufsize << 1)); 1321 ds->l_buf = ds->pending_buf + (ds->lit_bufsize << 2); 1322#else 1323 ds->sym_buf = ds->pending_buf + ds->lit_bufsize; 1324#endif 1325 1326 ds->l_desc.dyn_tree = ds->dyn_ltree; 1327 ds->d_desc.dyn_tree = ds->dyn_dtree; 1328 ds->bl_desc.dyn_tree = ds->bl_tree; 1329 1330 return Z_OK; 1331#endif /* MAXSEG_64K */ 1332} 1333 1334#ifndef FASTEST 1335/* =========================================================================== 1336 * Set match_start to the longest match starting at the given string and 1337 * return its length. Matches shorter or equal to prev_length are discarded, 1338 * in which case the result is equal to prev_length and match_start is 1339 * garbage. 1340 * IN assertions: cur_match is the head of the hash chain for the current 1341 * string (strstart) and its distance is <= MAX_DIST, and prev_length >= 1 1342 * OUT assertion: the match length is not greater than s->lookahead. 1343 */ 1344local uInt longest_match(deflate_state *s, IPos cur_match) { 1345 unsigned chain_length = s->max_chain_length;/* max hash chain length */ 1346 register Bytef *scan = s->window + s->strstart; /* current string */ 1347 register Bytef *match; /* matched string */ 1348 register int len; /* length of current match */ 1349 int best_len = (int)s->prev_length; /* best match length so far */ 1350 int nice_match = s->nice_match; /* stop if match long enough */ 1351 IPos limit = s->strstart > (IPos)MAX_DIST(s) ? 1352 s->strstart - (IPos)MAX_DIST(s) : NIL; 1353 /* Stop when cur_match becomes <= limit. To simplify the code, 1354 * we prevent matches with the string of window index 0. 1355 */ 1356 Posf *prev = s->prev; 1357 uInt wmask = s->w_mask; 1358 1359#ifdef UNALIGNED_OK 1360 /* Compare two bytes at a time. Note: this is not always beneficial. 1361 * Try with and without -DUNALIGNED_OK to check. 1362 */ 1363 register Bytef *strend = s->window + s->strstart + MAX_MATCH - 1; 1364 register ush scan_start = *(ushf*)scan; 1365 register ush scan_end = *(ushf*)(scan + best_len - 1); 1366#else 1367 register Bytef *strend = s->window + s->strstart + MAX_MATCH; 1368 register Byte scan_end1 = scan[best_len - 1]; 1369 register Byte scan_end = scan[best_len]; 1370#endif 1371 1372 /* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16. 1373 * It is easy to get rid of this optimization if necessary. 1374 */ 1375 Assert(s->hash_bits >= 8 && MAX_MATCH == 258, "Code too clever"); 1376 1377 /* Do not waste too much time if we already have a good match: */ 1378 if (s->prev_length >= s->good_match) { 1379 chain_length >>= 2; 1380 } 1381 /* Do not look for matches beyond the end of the input. This is necessary 1382 * to make deflate deterministic. 1383 */ 1384 if ((uInt)nice_match > s->lookahead) nice_match = (int)s->lookahead; 1385 1386 Assert((ulg)s->strstart <= s->window_size - MIN_LOOKAHEAD, 1387 "need lookahead"); 1388 1389 do { 1390 Assert(cur_match < s->strstart, "no future"); 1391 match = s->window + cur_match; 1392 1393 /* Skip to next match if the match length cannot increase 1394 * or if the match length is less than 2. Note that the checks below 1395 * for insufficient lookahead only occur occasionally for performance 1396 * reasons. Therefore uninitialized memory will be accessed, and 1397 * conditional jumps will be made that depend on those values. 1398 * However the length of the match is limited to the lookahead, so 1399 * the output of deflate is not affected by the uninitialized values. 1400 */ 1401#if (defined(UNALIGNED_OK) && MAX_MATCH == 258) 1402 /* This code assumes sizeof(unsigned short) == 2. Do not use 1403 * UNALIGNED_OK if your compiler uses a different size. 1404 */ 1405 if (*(ushf*)(match + best_len - 1) != scan_end || 1406 *(ushf*)match != scan_start) continue; 1407 1408 /* It is not necessary to compare scan[2] and match[2] since they are 1409 * always equal when the other bytes match, given that the hash keys 1410 * are equal and that HASH_BITS >= 8. Compare 2 bytes at a time at 1411 * strstart + 3, + 5, up to strstart + 257. We check for insufficient 1412 * lookahead only every 4th comparison; the 128th check will be made 1413 * at strstart + 257. If MAX_MATCH-2 is not a multiple of 8, it is 1414 * necessary to put more guard bytes at the end of the window, or 1415 * to check more often for insufficient lookahead. 1416 */ 1417 Assert(scan[2] == match[2], "scan[2]?"); 1418 scan++, match++; 1419 do { 1420 } while (*(ushf*)(scan += 2) == *(ushf*)(match += 2) && 1421 *(ushf*)(scan += 2) == *(ushf*)(match += 2) && 1422 *(ushf*)(scan += 2) == *(ushf*)(match += 2) && 1423 *(ushf*)(scan += 2) == *(ushf*)(match += 2) && 1424 scan < strend); 1425 /* The funny "do {}" generates better code on most compilers */ 1426 1427 /* Here, scan <= window + strstart + 257 */ 1428 Assert(scan <= s->window + (unsigned)(s->window_size - 1), 1429 "wild scan"); 1430 if (*scan == *match) scan++; 1431 1432 len = (MAX_MATCH - 1) - (int)(strend - scan); 1433 scan = strend - (MAX_MATCH-1); 1434 1435#else /* UNALIGNED_OK */ 1436 1437 if (match[best_len] != scan_end || 1438 match[best_len - 1] != scan_end1 || 1439 *match != *scan || 1440 *++match != scan[1]) continue; 1441 1442 /* The check at best_len - 1 can be removed because it will be made 1443 * again later. (This heuristic is not always a win.) 1444 * It is not necessary to compare scan[2] and match[2] since they 1445 * are always equal when the other bytes match, given that 1446 * the hash keys are equal and that HASH_BITS >= 8. 1447 */ 1448 scan += 2, match++; 1449 Assert(*scan == *match, "match[2]?"); 1450 1451 /* We check for insufficient lookahead only every 8th comparison; 1452 * the 256th check will be made at strstart + 258. 1453 */ 1454 do { 1455 } while (*++scan == *++match && *++scan == *++match && 1456 *++scan == *++match && *++scan == *++match && 1457 *++scan == *++match && *++scan == *++match && 1458 *++scan == *++match && *++scan == *++match && 1459 scan < strend); 1460 1461 Assert(scan <= s->window + (unsigned)(s->window_size - 1), 1462 "wild scan"); 1463 1464 len = MAX_MATCH - (int)(strend - scan); 1465 scan = strend - MAX_MATCH; 1466 1467#endif /* UNALIGNED_OK */ 1468 1469 if (len > best_len) { 1470 s->match_start = cur_match; 1471 best_len = len; 1472 if (len >= nice_match) break; 1473#ifdef UNALIGNED_OK 1474 scan_end = *(ushf*)(scan + best_len - 1); 1475#else 1476 scan_end1 = scan[best_len - 1]; 1477 scan_end = scan[best_len]; 1478#endif 1479 } 1480 } while ((cur_match = prev[cur_match & wmask]) > limit 1481 && --chain_length != 0); 1482 1483 if ((uInt)best_len <= s->lookahead) return (uInt)best_len; 1484 return s->lookahead; 1485} 1486 1487#else /* FASTEST */ 1488 1489/* --------------------------------------------------------------------------- 1490 * Optimized version for FASTEST only 1491 */ 1492local uInt longest_match(deflate_state *s, IPos cur_match) { 1493 register Bytef *scan = s->window + s->strstart; /* current string */ 1494 register Bytef *match; /* matched string */ 1495 register int len; /* length of current match */ 1496 register Bytef *strend = s->window + s->strstart + MAX_MATCH; 1497 1498 /* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16. 1499 * It is easy to get rid of this optimization if necessary. 1500 */ 1501 Assert(s->hash_bits >= 8 && MAX_MATCH == 258, "Code too clever"); 1502 1503 Assert((ulg)s->strstart <= s->window_size - MIN_LOOKAHEAD, 1504 "need lookahead"); 1505 1506 Assert(cur_match < s->strstart, "no future"); 1507 1508 match = s->window + cur_match; 1509 1510 /* Return failure if the match length is less than 2: 1511 */ 1512 if (match[0] != scan[0] || match[1] != scan[1]) return MIN_MATCH-1; 1513 1514 /* The check at best_len - 1 can be removed because it will be made 1515 * again later. (This heuristic is not always a win.) 1516 * It is not necessary to compare scan[2] and match[2] since they 1517 * are always equal when the other bytes match, given that 1518 * the hash keys are equal and that HASH_BITS >= 8. 1519 */ 1520 scan += 2, match += 2; 1521 Assert(*scan == *match, "match[2]?"); 1522 1523 /* We check for insufficient lookahead only every 8th comparison; 1524 * the 256th check will be made at strstart + 258. 1525 */ 1526 do { 1527 } while (*++scan == *++match && *++scan == *++match && 1528 *++scan == *++match && *++scan == *++match && 1529 *++scan == *++match && *++scan == *++match && 1530 *++scan == *++match && *++scan == *++match && 1531 scan < strend); 1532 1533 Assert(scan <= s->window + (unsigned)(s->window_size - 1), "wild scan"); 1534 1535 len = MAX_MATCH - (int)(strend - scan); 1536 1537 if (len < MIN_MATCH) return MIN_MATCH - 1; 1538 1539 s->match_start = cur_match; 1540 return (uInt)len <= s->lookahead ? (uInt)len : s->lookahead; 1541} 1542 1543#endif /* FASTEST */ 1544 1545#ifdef ZLIB_DEBUG 1546 1547#define EQUAL 0 1548/* result of memcmp for equal strings */ 1549 1550/* =========================================================================== 1551 * Check that the match at match_start is indeed a match. 1552 */ 1553local void check_match(deflate_state *s, IPos start, IPos match, int length) { 1554 /* check that the match is indeed a match */ 1555 Bytef *back = s->window + (int)match, *here = s->window + start; 1556 IPos len = length; 1557 if (match == (IPos)-1) { 1558 /* match starts one byte before the current window -- just compare the 1559 subsequent length-1 bytes */ 1560 back++; 1561 here++; 1562 len--; 1563 } 1564 if (zmemcmp(back, here, len) != EQUAL) { 1565 fprintf(stderr, " start %u, match %d, length %d\n", 1566 start, (int)match, length); 1567 do { 1568 fprintf(stderr, "(%02x %02x)", *back++, *here++); 1569 } while (--len != 0); 1570 z_error("invalid match"); 1571 } 1572 if (z_verbose > 1) { 1573 fprintf(stderr,"\\[%d,%d]", start - match, length); 1574 do { putc(s->window[start++], stderr); } while (--length != 0); 1575 } 1576} 1577#else 1578# define check_match(s, start, match, length) 1579#endif /* ZLIB_DEBUG */ 1580 1581/* =========================================================================== 1582 * Flush the current block, with given end-of-file flag. 1583 * IN assertion: strstart is set to the end of the current match. 1584 */ 1585#define FLUSH_BLOCK_ONLY(s, last) { \ 1586 _tr_flush_block(s, (s->block_start >= 0L ? \ 1587 (charf *)&s->window[(unsigned)s->block_start] : \ 1588 (charf *)Z_NULL), \ 1589 (ulg)((long)s->strstart - s->block_start), \ 1590 (last)); \ 1591 s->block_start = s->strstart; \ 1592 flush_pending(s->strm); \ 1593 Tracev((stderr,"[FLUSH]")); \ 1594} 1595 1596/* Same but force premature exit if necessary. */ 1597#define FLUSH_BLOCK(s, last) { \ 1598 FLUSH_BLOCK_ONLY(s, last); \ 1599 if (s->strm->avail_out == 0) return (last) ? finish_started : need_more; \ 1600} 1601 1602/* Maximum stored block length in deflate format (not including header). */ 1603#define MAX_STORED 65535 1604 1605/* Minimum of a and b. */ 1606#define MIN(a, b) ((a) > (b) ? (b) : (a)) 1607 1608/* =========================================================================== 1609 * Copy without compression as much as possible from the input stream, return 1610 * the current block state. 1611 * 1612 * In case deflateParams() is used to later switch to a non-zero compression 1613 * level, s->matches (otherwise unused when storing) keeps track of the number 1614 * of hash table slides to perform. If s->matches is 1, then one hash table 1615 * slide will be done when switching. If s->matches is 2, the maximum value 1616 * allowed here, then the hash table will be cleared, since two or more slides 1617 * is the same as a clear. 1618 * 1619 * deflate_stored() is written to minimize the number of times an input byte is 1620 * copied. It is most efficient with large input and output buffers, which 1621 * maximizes the opportunities to have a single copy from next_in to next_out. 1622 */ 1623local block_state deflate_stored(deflate_state *s, int flush) { 1624 /* Smallest worthy block size when not flushing or finishing. By default 1625 * this is 32K. This can be as small as 507 bytes for memLevel == 1. For 1626 * large input and output buffers, the stored block size will be larger. 1627 */ 1628 unsigned min_block = MIN(s->pending_buf_size - 5, s->w_size); 1629 1630 /* Copy as many min_block or larger stored blocks directly to next_out as 1631 * possible. If flushing, copy the remaining available input to next_out as 1632 * stored blocks, if there is enough space. 1633 */ 1634 int last = 0; 1635 unsigned len, left, have; 1636 unsigned used = s->strm->avail_in; 1637 do { 1638 /* Set len to the maximum size block that we can copy directly with the 1639 * available input data and output space. Set left to how much of that 1640 * would be copied from what's left in the window. 1641 */ 1642 len = MAX_STORED; /* maximum deflate stored block length */ 1643 have = (s->bi_valid + 42) >> 3; /* number of header bytes */ 1644 if (s->strm->avail_out < have) /* need room for header */ 1645 break; 1646 /* maximum stored block length that will fit in avail_out: */ 1647 have = s->strm->avail_out - have; 1648 left = s->strstart - s->block_start; /* bytes left in window */ 1649 if (len > (ulg)left + s->strm->avail_in) 1650 len = left + s->strm->avail_in; /* limit len to the input */ 1651 if (len > have) 1652 len = have; /* limit len to the output */ 1653 1654 /* If the stored block would be less than min_block in length, or if 1655 * unable to copy all of the available input when flushing, then try 1656 * copying to the window and the pending buffer instead. Also don't 1657 * write an empty block when flushing -- deflate() does that. 1658 */ 1659 if (len < min_block && ((len == 0 && flush != Z_FINISH) || 1660 flush == Z_NO_FLUSH || 1661 len != left + s->strm->avail_in)) 1662 break; 1663 1664 /* Make a dummy stored block in pending to get the header bytes, 1665 * including any pending bits. This also updates the debugging counts. 1666 */ 1667 last = flush == Z_FINISH && len == left + s->strm->avail_in ? 1 : 0; 1668 _tr_stored_block(s, (char *)0, 0L, last); 1669 1670 /* Replace the lengths in the dummy stored block with len. */ 1671 s->pending_buf[s->pending - 4] = (Bytef)len; 1672 s->pending_buf[s->pending - 3] = (Bytef)(len >> 8); 1673 s->pending_buf[s->pending - 2] = (Bytef)~len; 1674 s->pending_buf[s->pending - 1] = (Bytef)(~len >> 8); 1675 1676 /* Write the stored block header bytes. */ 1677 flush_pending(s->strm); 1678 1679#ifdef ZLIB_DEBUG 1680 /* Update debugging counts for the data about to be copied. */ 1681 s->compressed_len += len << 3; 1682 s->bits_sent += len << 3; 1683#endif 1684 1685 /* Copy uncompressed bytes from the window to next_out. */ 1686 if (left) { 1687 if (left > len) 1688 left = len; 1689 zmemcpy(s->strm->next_out, s->window + s->block_start, left); 1690 s->strm->next_out += left; 1691 s->strm->avail_out -= left; 1692 s->strm->total_out += left; 1693 s->block_start += left; 1694 len -= left; 1695 } 1696 1697 /* Copy uncompressed bytes directly from next_in to next_out, updating 1698 * the check value. 1699 */ 1700 if (len) { 1701 read_buf(s->strm, s->strm->next_out, len); 1702 s->strm->next_out += len; 1703 s->strm->avail_out -= len; 1704 s->strm->total_out += len; 1705 } 1706 } while (last == 0); 1707 1708 /* Update the sliding window with the last s->w_size bytes of the copied 1709 * data, or append all of the copied data to the existing window if less 1710 * than s->w_size bytes were copied. Also update the number of bytes to 1711 * insert in the hash tables, in the event that deflateParams() switches to 1712 * a non-zero compression level. 1713 */ 1714 used -= s->strm->avail_in; /* number of input bytes directly copied */ 1715 if (used) { 1716 /* If any input was used, then no unused input remains in the window, 1717 * therefore s->block_start == s->strstart. 1718 */ 1719 if (used >= s->w_size) { /* supplant the previous history */ 1720 s->matches = 2; /* clear hash */ 1721 zmemcpy(s->window, s->strm->next_in - s->w_size, s->w_size); 1722 s->strstart = s->w_size; 1723 s->insert = s->strstart; 1724 } 1725 else { 1726 if (s->window_size - s->strstart <= used) { 1727 /* Slide the window down. */ 1728 s->strstart -= s->w_size; 1729 zmemcpy(s->window, s->window + s->w_size, s->strstart); 1730 if (s->matches < 2) 1731 s->matches++; /* add a pending slide_hash() */ 1732 if (s->insert > s->strstart) 1733 s->insert = s->strstart; 1734 } 1735 zmemcpy(s->window + s->strstart, s->strm->next_in - used, used); 1736 s->strstart += used; 1737 s->insert += MIN(used, s->w_size - s->insert); 1738 } 1739 s->block_start = s->strstart; 1740 } 1741 if (s->high_water < s->strstart) 1742 s->high_water = s->strstart; 1743 1744 /* If the last block was written to next_out, then done. */ 1745 if (last) 1746 return finish_done; 1747 1748 /* If flushing and all input has been consumed, then done. */ 1749 if (flush != Z_NO_FLUSH && flush != Z_FINISH && 1750 s->strm->avail_in == 0 && (long)s->strstart == s->block_start) 1751 return block_done; 1752 1753 /* Fill the window with any remaining input. */ 1754 have = s->window_size - s->strstart; 1755 if (s->strm->avail_in > have && s->block_start >= (long)s->w_size) { 1756 /* Slide the window down. */ 1757 s->block_start -= s->w_size; 1758 s->strstart -= s->w_size; 1759 zmemcpy(s->window, s->window + s->w_size, s->strstart); 1760 if (s->matches < 2) 1761 s->matches++; /* add a pending slide_hash() */ 1762 have += s->w_size; /* more space now */ 1763 if (s->insert > s->strstart) 1764 s->insert = s->strstart; 1765 } 1766 if (have > s->strm->avail_in) 1767 have = s->strm->avail_in; 1768 if (have) { 1769 read_buf(s->strm, s->window + s->strstart, have); 1770 s->strstart += have; 1771 s->insert += MIN(have, s->w_size - s->insert); 1772 } 1773 if (s->high_water < s->strstart) 1774 s->high_water = s->strstart; 1775 1776 /* There was not enough avail_out to write a complete worthy or flushed 1777 * stored block to next_out. Write a stored block to pending instead, if we 1778 * have enough input for a worthy block, or if flushing and there is enough 1779 * room for the remaining input as a stored block in the pending buffer. 1780 */ 1781 have = (s->bi_valid + 42) >> 3; /* number of header bytes */ 1782 /* maximum stored block length that will fit in pending: */ 1783 have = MIN(s->pending_buf_size - have, MAX_STORED); 1784 min_block = MIN(have, s->w_size); 1785 left = s->strstart - s->block_start; 1786 if (left >= min_block || 1787 ((left || flush == Z_FINISH) && flush != Z_NO_FLUSH && 1788 s->strm->avail_in == 0 && left <= have)) { 1789 len = MIN(left, have); 1790 last = flush == Z_FINISH && s->strm->avail_in == 0 && 1791 len == left ? 1 : 0; 1792 _tr_stored_block(s, (charf *)s->window + s->block_start, len, last); 1793 s->block_start += len; 1794 flush_pending(s->strm); 1795 } 1796 1797 /* We've done all we can with the available input and output. */ 1798 return last ? finish_started : need_more; 1799} 1800 1801/* =========================================================================== 1802 * Compress as much as possible from the input stream, return the current 1803 * block state. 1804 * This function does not perform lazy evaluation of matches and inserts 1805 * new strings in the dictionary only for unmatched strings or for short 1806 * matches. It is used only for the fast compression options. 1807 */ 1808local block_state deflate_fast(deflate_state *s, int flush) { 1809 IPos hash_head; /* head of the hash chain */ 1810 int bflush; /* set if current block must be flushed */ 1811 1812 for (;;) { 1813 /* Make sure that we always have enough lookahead, except 1814 * at the end of the input file. We need MAX_MATCH bytes 1815 * for the next match, plus MIN_MATCH bytes to insert the 1816 * string following the next match. 1817 */ 1818 if (s->lookahead < MIN_LOOKAHEAD) { 1819 fill_window(s); 1820 if (s->lookahead < MIN_LOOKAHEAD && flush == Z_NO_FLUSH) { 1821 return need_more; 1822 } 1823 if (s->lookahead == 0) break; /* flush the current block */ 1824 } 1825 1826 /* Insert the string window[strstart .. strstart + 2] in the 1827 * dictionary, and set hash_head to the head of the hash chain: 1828 */ 1829 hash_head = NIL; 1830 if (s->lookahead >= MIN_MATCH) { 1831 INSERT_STRING(s, s->strstart, hash_head); 1832 } 1833 1834 /* Find the longest match, discarding those <= prev_length. 1835 * At this point we have always match_length < MIN_MATCH 1836 */ 1837 if (hash_head != NIL && s->strstart - hash_head <= MAX_DIST(s)) { 1838 /* To simplify the code, we prevent matches with the string 1839 * of window index 0 (in particular we have to avoid a match 1840 * of the string with itself at the start of the input file). 1841 */ 1842 s->match_length = longest_match (s, hash_head); 1843 /* longest_match() sets match_start */ 1844 } 1845 if (s->match_length >= MIN_MATCH) { 1846 check_match(s, s->strstart, s->match_start, s->match_length); 1847 1848 _tr_tally_dist(s, s->strstart - s->match_start, 1849 s->match_length - MIN_MATCH, bflush); 1850 1851 s->lookahead -= s->match_length; 1852 1853 /* Insert new strings in the hash table only if the match length 1854 * is not too large. This saves time but degrades compression. 1855 */ 1856#ifndef FASTEST 1857 if (s->match_length <= s->max_insert_length && 1858 s->lookahead >= MIN_MATCH) { 1859 s->match_length--; /* string at strstart already in table */ 1860 do { 1861 s->strstart++; 1862 INSERT_STRING(s, s->strstart, hash_head); 1863 /* strstart never exceeds WSIZE-MAX_MATCH, so there are 1864 * always MIN_MATCH bytes ahead. 1865 */ 1866 } while (--s->match_length != 0); 1867 s->strstart++; 1868 } else 1869#endif 1870 { 1871 s->strstart += s->match_length; 1872 s->match_length = 0; 1873 s->ins_h = s->window[s->strstart]; 1874 UPDATE_HASH(s, s->ins_h, s->window[s->strstart + 1]); 1875#if MIN_MATCH != 3 1876 Call UPDATE_HASH() MIN_MATCH-3 more times 1877#endif 1878 /* If lookahead < MIN_MATCH, ins_h is garbage, but it does not 1879 * matter since it will be recomputed at next deflate call. 1880 */ 1881 } 1882 } else { 1883 /* No match, output a literal byte */ 1884 Tracevv((stderr,"%c", s->window[s->strstart])); 1885 _tr_tally_lit(s, s->window[s->strstart], bflush); 1886 s->lookahead--; 1887 s->strstart++; 1888 } 1889 if (bflush) FLUSH_BLOCK(s, 0); 1890 } 1891 s->insert = s->strstart < MIN_MATCH-1 ? s->strstart : MIN_MATCH-1; 1892 if (flush == Z_FINISH) { 1893 FLUSH_BLOCK(s, 1); 1894 return finish_done; 1895 } 1896 if (s->sym_next) 1897 FLUSH_BLOCK(s, 0); 1898 return block_done; 1899} 1900 1901#ifndef FASTEST 1902/* =========================================================================== 1903 * Same as above, but achieves better compression. We use a lazy 1904 * evaluation for matches: a match is finally adopted only if there is 1905 * no better match at the next window position. 1906 */ 1907local block_state deflate_slow(deflate_state *s, int flush) { 1908 IPos hash_head; /* head of hash chain */ 1909 int bflush; /* set if current block must be flushed */ 1910 1911 /* Process the input block. */ 1912 for (;;) { 1913 /* Make sure that we always have enough lookahead, except 1914 * at the end of the input file. We need MAX_MATCH bytes 1915 * for the next match, plus MIN_MATCH bytes to insert the 1916 * string following the next match. 1917 */ 1918 if (s->lookahead < MIN_LOOKAHEAD) { 1919 fill_window(s); 1920 if (s->lookahead < MIN_LOOKAHEAD && flush == Z_NO_FLUSH) { 1921 return need_more; 1922 } 1923 if (s->lookahead == 0) break; /* flush the current block */ 1924 } 1925 1926 /* Insert the string window[strstart .. strstart + 2] in the 1927 * dictionary, and set hash_head to the head of the hash chain: 1928 */ 1929 hash_head = NIL; 1930 if (s->lookahead >= MIN_MATCH) { 1931 INSERT_STRING(s, s->strstart, hash_head); 1932 } 1933 1934 /* Find the longest match, discarding those <= prev_length. 1935 */ 1936 s->prev_length = s->match_length, s->prev_match = s->match_start; 1937 s->match_length = MIN_MATCH-1; 1938 1939 if (hash_head != NIL && s->prev_length < s->max_lazy_match && 1940 s->strstart - hash_head <= MAX_DIST(s)) { 1941 /* To simplify the code, we prevent matches with the string 1942 * of window index 0 (in particular we have to avoid a match 1943 * of the string with itself at the start of the input file). 1944 */ 1945 s->match_length = longest_match (s, hash_head); 1946 /* longest_match() sets match_start */ 1947 1948 if (s->match_length <= 5 && (s->strategy == Z_FILTERED 1949#if TOO_FAR <= 32767 1950 || (s->match_length == MIN_MATCH && 1951 s->strstart - s->match_start > TOO_FAR) 1952#endif 1953 )) { 1954 1955 /* If prev_match is also MIN_MATCH, match_start is garbage 1956 * but we will ignore the current match anyway. 1957 */ 1958 s->match_length = MIN_MATCH-1; 1959 } 1960 } 1961 /* If there was a match at the previous step and the current 1962 * match is not better, output the previous match: 1963 */ 1964 if (s->prev_length >= MIN_MATCH && s->match_length <= s->prev_length) { 1965 uInt max_insert = s->strstart + s->lookahead - MIN_MATCH; 1966 /* Do not insert strings in hash table beyond this. */ 1967 1968 check_match(s, s->strstart - 1, s->prev_match, s->prev_length); 1969 1970 _tr_tally_dist(s, s->strstart - 1 - s->prev_match, 1971 s->prev_length - MIN_MATCH, bflush); 1972 1973 /* Insert in hash table all strings up to the end of the match. 1974 * strstart - 1 and strstart are already inserted. If there is not 1975 * enough lookahead, the last two strings are not inserted in 1976 * the hash table. 1977 */ 1978 s->lookahead -= s->prev_length - 1; 1979 s->prev_length -= 2; 1980 do { 1981 if (++s->strstart <= max_insert) { 1982 INSERT_STRING(s, s->strstart, hash_head); 1983 } 1984 } while (--s->prev_length != 0); 1985 s->match_available = 0; 1986 s->match_length = MIN_MATCH-1; 1987 s->strstart++; 1988 1989 if (bflush) FLUSH_BLOCK(s, 0); 1990 1991 } else if (s->match_available) { 1992 /* If there was no match at the previous position, output a 1993 * single literal. If there was a match but the current match 1994 * is longer, truncate the previous match to a single literal. 1995 */ 1996 Tracevv((stderr,"%c", s->window[s->strstart - 1])); 1997 _tr_tally_lit(s, s->window[s->strstart - 1], bflush); 1998 if (bflush) { 1999 FLUSH_BLOCK_ONLY(s, 0); 2000 } 2001 s->strstart++; 2002 s->lookahead--; 2003 if (s->strm->avail_out == 0) return need_more; 2004 } else { 2005 /* There is no previous match to compare with, wait for 2006 * the next step to decide. 2007 */ 2008 s->match_available = 1; 2009 s->strstart++; 2010 s->lookahead--; 2011 } 2012 } 2013 Assert (flush != Z_NO_FLUSH, "no flush?"); 2014 if (s->match_available) { 2015 Tracevv((stderr,"%c", s->window[s->strstart - 1])); 2016 _tr_tally_lit(s, s->window[s->strstart - 1], bflush); 2017 s->match_available = 0; 2018 } 2019 s->insert = s->strstart < MIN_MATCH-1 ? s->strstart : MIN_MATCH-1; 2020 if (flush == Z_FINISH) { 2021 FLUSH_BLOCK(s, 1); 2022 return finish_done; 2023 } 2024 if (s->sym_next) 2025 FLUSH_BLOCK(s, 0); 2026 return block_done; 2027} 2028#endif /* FASTEST */ 2029 2030/* =========================================================================== 2031 * For Z_RLE, simply look for runs of bytes, generate matches only of distance 2032 * one. Do not maintain a hash table. (It will be regenerated if this run of 2033 * deflate switches away from Z_RLE.) 2034 */ 2035local block_state deflate_rle(deflate_state *s, int flush) { 2036 int bflush; /* set if current block must be flushed */ 2037 uInt prev; /* byte at distance one to match */ 2038 Bytef *scan, *strend; /* scan goes up to strend for length of run */ 2039 2040 for (;;) { 2041 /* Make sure that we always have enough lookahead, except 2042 * at the end of the input file. We need MAX_MATCH bytes 2043 * for the longest run, plus one for the unrolled loop. 2044 */ 2045 if (s->lookahead <= MAX_MATCH) { 2046 fill_window(s); 2047 if (s->lookahead <= MAX_MATCH && flush == Z_NO_FLUSH) { 2048 return need_more; 2049 } 2050 if (s->lookahead == 0) break; /* flush the current block */ 2051 } 2052 2053 /* See how many times the previous byte repeats */ 2054 s->match_length = 0; 2055 if (s->lookahead >= MIN_MATCH && s->strstart > 0) { 2056 scan = s->window + s->strstart - 1; 2057 prev = *scan; 2058 if (prev == *++scan && prev == *++scan && prev == *++scan) { 2059 strend = s->window + s->strstart + MAX_MATCH; 2060 do { 2061 } while (prev == *++scan && prev == *++scan && 2062 prev == *++scan && prev == *++scan && 2063 prev == *++scan && prev == *++scan && 2064 prev == *++scan && prev == *++scan && 2065 scan < strend); 2066 s->match_length = MAX_MATCH - (uInt)(strend - scan); 2067 if (s->match_length > s->lookahead) 2068 s->match_length = s->lookahead; 2069 } 2070 Assert(scan <= s->window + (uInt)(s->window_size - 1), 2071 "wild scan"); 2072 } 2073 2074 /* Emit match if have run of MIN_MATCH or longer, else emit literal */ 2075 if (s->match_length >= MIN_MATCH) { 2076 check_match(s, s->strstart, s->strstart - 1, s->match_length); 2077 2078 _tr_tally_dist(s, 1, s->match_length - MIN_MATCH, bflush); 2079 2080 s->lookahead -= s->match_length; 2081 s->strstart += s->match_length; 2082 s->match_length = 0; 2083 } else { 2084 /* No match, output a literal byte */ 2085 Tracevv((stderr,"%c", s->window[s->strstart])); 2086 _tr_tally_lit(s, s->window[s->strstart], bflush); 2087 s->lookahead--; 2088 s->strstart++; 2089 } 2090 if (bflush) FLUSH_BLOCK(s, 0); 2091 } 2092 s->insert = 0; 2093 if (flush == Z_FINISH) { 2094 FLUSH_BLOCK(s, 1); 2095 return finish_done; 2096 } 2097 if (s->sym_next) 2098 FLUSH_BLOCK(s, 0); 2099 return block_done; 2100} 2101 2102/* =========================================================================== 2103 * For Z_HUFFMAN_ONLY, do not look for matches. Do not maintain a hash table. 2104 * (It will be regenerated if this run of deflate switches away from Huffman.) 2105 */ 2106local block_state deflate_huff(deflate_state *s, int flush) { 2107 int bflush; /* set if current block must be flushed */ 2108 2109 for (;;) { 2110 /* Make sure that we have a literal to write. */ 2111 if (s->lookahead == 0) { 2112 fill_window(s); 2113 if (s->lookahead == 0) { 2114 if (flush == Z_NO_FLUSH) 2115 return need_more; 2116 break; /* flush the current block */ 2117 } 2118 } 2119 2120 /* Output a literal byte */ 2121 s->match_length = 0; 2122 Tracevv((stderr,"%c", s->window[s->strstart])); 2123 _tr_tally_lit(s, s->window[s->strstart], bflush); 2124 s->lookahead--; 2125 s->strstart++; 2126 if (bflush) FLUSH_BLOCK(s, 0); 2127 } 2128 s->insert = 0; 2129 if (flush == Z_FINISH) { 2130 FLUSH_BLOCK(s, 1); 2131 return finish_done; 2132 } 2133 if (s->sym_next) 2134 FLUSH_BLOCK(s, 0); 2135 return block_done; 2136} 2137