1 /* deflate.c -- compress data using the deflation algorithm 2 * Copyright (C) 1995-2022 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 /* @(#) $Id$ */ 51 52 #include "deflate.h" 53 54 const char deflate_copyright[] = 55 " deflate 1.2.12 Copyright 1995-2022 Jean-loup Gailly and Mark Adler "; 56 /* 57 If you use the zlib library in a product, an acknowledgment is welcome 58 in the documentation of your product. If for some reason you cannot 59 include such an acknowledgment, I would appreciate that you keep this 60 copyright string in the executable of your product. 61 */ 62 63 /* =========================================================================== 64 * Function prototypes. 65 */ 66 typedef enum { 67 need_more, /* block not completed, need more input or more output */ 68 block_done, /* block flush performed */ 69 finish_started, /* finish started, need only more output at next deflate */ 70 finish_done /* finish done, accept no more input or output */ 71 } block_state; 72 73 typedef block_state (*compress_func) OF((deflate_state *s, int flush)); 74 /* Compression function. Returns the block state after the call. */ 75 76 local int deflateStateCheck OF((z_streamp strm)); 77 local void slide_hash OF((deflate_state *s)); 78 local void fill_window OF((deflate_state *s)); 79 local block_state deflate_stored OF((deflate_state *s, int flush)); 80 local block_state deflate_fast OF((deflate_state *s, int flush)); 81 #ifndef FASTEST 82 local block_state deflate_slow OF((deflate_state *s, int flush)); 83 #endif 84 local block_state deflate_rle OF((deflate_state *s, int flush)); 85 local block_state deflate_huff OF((deflate_state *s, int flush)); 86 local void lm_init OF((deflate_state *s)); 87 local void putShortMSB OF((deflate_state *s, uInt b)); 88 local void flush_pending OF((z_streamp strm)); 89 local unsigned read_buf OF((z_streamp strm, Bytef *buf, unsigned size)); 90 #ifdef ASMV 91 # pragma message("Assembler code may have bugs -- use at your own risk") 92 void match_init OF((void)); /* asm code initialization */ 93 uInt longest_match OF((deflate_state *s, IPos cur_match)); 94 #else 95 local uInt longest_match OF((deflate_state *s, IPos cur_match)); 96 #endif 97 98 #ifdef ZLIB_DEBUG 99 local void check_match OF((deflate_state *s, IPos start, IPos match, 100 int length)); 101 #endif 102 103 /* =========================================================================== 104 * Local data 105 */ 106 107 #define NIL 0 108 /* Tail of hash chains */ 109 110 #ifndef TOO_FAR 111 # define TOO_FAR 4096 112 #endif 113 /* Matches of length 3 are discarded if their distance exceeds TOO_FAR */ 114 115 /* Values for max_lazy_match, good_match and max_chain_length, depending on 116 * the desired pack level (0..9). The values given below have been tuned to 117 * exclude worst case performance for pathological files. Better values may be 118 * found for specific files. 119 */ 120 typedef struct config_s { 121 ush good_length; /* reduce lazy search above this match length */ 122 ush max_lazy; /* do not perform lazy search above this match length */ 123 ush nice_length; /* quit search above this match length */ 124 ush max_chain; 125 compress_func func; 126 } config; 127 128 #ifdef FASTEST 129 local const config configuration_table[2] = { 130 /* good lazy nice chain */ 131 /* 0 */ {0, 0, 0, 0, deflate_stored}, /* store only */ 132 /* 1 */ {4, 4, 8, 4, deflate_fast}}; /* max speed, no lazy matches */ 133 #else 134 local const config configuration_table[10] = { 135 /* good lazy nice chain */ 136 /* 0 */ {0, 0, 0, 0, deflate_stored}, /* store only */ 137 /* 1 */ {4, 4, 8, 4, deflate_fast}, /* max speed, no lazy matches */ 138 /* 2 */ {4, 5, 16, 8, deflate_fast}, 139 /* 3 */ {4, 6, 32, 32, deflate_fast}, 140 141 /* 4 */ {4, 4, 16, 16, deflate_slow}, /* lazy matches */ 142 /* 5 */ {8, 16, 32, 32, deflate_slow}, 143 /* 6 */ {8, 16, 128, 128, deflate_slow}, 144 /* 7 */ {8, 32, 128, 256, deflate_slow}, 145 /* 8 */ {32, 128, 258, 1024, deflate_slow}, 146 /* 9 */ {32, 258, 258, 4096, deflate_slow}}; /* max compression */ 147 #endif 148 149 /* Note: the deflate() code requires max_lazy >= MIN_MATCH and max_chain >= 4 150 * For deflate_fast() (levels <= 3) good is ignored and lazy has a different 151 * meaning. 152 */ 153 154 /* rank Z_BLOCK between Z_NO_FLUSH and Z_PARTIAL_FLUSH */ 155 #define RANK(f) (((f) * 2) - ((f) > 4 ? 9 : 0)) 156 157 /* =========================================================================== 158 * Update a hash value with the given input byte 159 * IN assertion: all calls to UPDATE_HASH are made with consecutive input 160 * characters, so that a running hash key can be computed from the previous 161 * key instead of complete recalculation each time. 162 */ 163 #define UPDATE_HASH(s,h,c) (h = (((h)<<s->hash_shift) ^ (c)) & s->hash_mask) 164 165 166 /* =========================================================================== 167 * Insert string str in the dictionary and set match_head to the previous head 168 * of the hash chain (the most recent string with same hash key). Return 169 * the previous length of the hash chain. 170 * If this file is compiled with -DFASTEST, the compression level is forced 171 * to 1, and no hash chains are maintained. 172 * IN assertion: all calls to INSERT_STRING are made with consecutive input 173 * characters and the first MIN_MATCH bytes of str are valid (except for 174 * the last MIN_MATCH-1 bytes of the input file). 175 */ 176 #ifdef FASTEST 177 #define INSERT_STRING(s, str, match_head) \ 178 (UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \ 179 match_head = s->head[s->ins_h], \ 180 s->head[s->ins_h] = (Pos)(str)) 181 #else 182 #define INSERT_STRING(s, str, match_head) \ 183 (UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \ 184 match_head = s->prev[(str) & s->w_mask] = s->head[s->ins_h], \ 185 s->head[s->ins_h] = (Pos)(str)) 186 #endif 187 188 /* =========================================================================== 189 * Initialize the hash table (avoiding 64K overflow for 16 bit systems). 190 * prev[] will be initialized on the fly. 191 */ 192 #define CLEAR_HASH(s) \ 193 do { \ 194 s->head[s->hash_size-1] = NIL; \ 195 zmemzero((Bytef *)s->head, \ 196 (unsigned)(s->hash_size-1)*sizeof(*s->head)); \ 197 } while (0) 198 199 /* =========================================================================== 200 * Slide the hash table when sliding the window down (could be avoided with 32 201 * bit values at the expense of memory usage). We slide even when level == 0 to 202 * keep the hash table consistent if we switch back to level > 0 later. 203 */ 204 local void slide_hash(s) 205 deflate_state *s; 206 { 207 unsigned n, m; 208 Posf *p; 209 uInt wsize = s->w_size; 210 211 n = s->hash_size; 212 p = &s->head[n]; 213 do { 214 m = *--p; 215 *p = (Pos)(m >= wsize ? m - wsize : NIL); 216 } while (--n); 217 n = wsize; 218 #ifndef FASTEST 219 p = &s->prev[n]; 220 do { 221 m = *--p; 222 *p = (Pos)(m >= wsize ? m - wsize : NIL); 223 /* If n is not on any hash chain, prev[n] is garbage but 224 * its value will never be used. 225 */ 226 } while (--n); 227 #endif 228 } 229 230 /* ========================================================================= */ 231 int ZEXPORT deflateInit_(strm, level, version, stream_size) 232 z_streamp strm; 233 int level; 234 const char *version; 235 int stream_size; 236 { 237 return deflateInit2_(strm, level, Z_DEFLATED, MAX_WBITS, DEF_MEM_LEVEL, 238 Z_DEFAULT_STRATEGY, version, stream_size); 239 /* To do: ignore strm->next_in if we use it as window */ 240 } 241 242 /* ========================================================================= */ 243 int ZEXPORT deflateInit2_(strm, level, method, windowBits, memLevel, strategy, 244 version, stream_size) 245 z_streamp strm; 246 int level; 247 int method; 248 int windowBits; 249 int memLevel; 250 int strategy; 251 const char *version; 252 int stream_size; 253 { 254 deflate_state *s; 255 int wrap = 1; 256 static const char my_version[] = ZLIB_VERSION; 257 258 if (version == Z_NULL || version[0] != my_version[0] || 259 stream_size != sizeof(z_stream)) { 260 return Z_VERSION_ERROR; 261 } 262 if (strm == Z_NULL) return Z_STREAM_ERROR; 263 264 strm->msg = Z_NULL; 265 if (strm->zalloc == (alloc_func)0) { 266 #if defined(Z_SOLO) && !defined(_KERNEL) 267 return Z_STREAM_ERROR; 268 #else 269 strm->zalloc = zcalloc; 270 strm->opaque = (voidpf)0; 271 #endif 272 } 273 if (strm->zfree == (free_func)0) 274 #if defined(Z_SOLO) && !defined(_KERNEL) 275 return Z_STREAM_ERROR; 276 #else 277 strm->zfree = zcfree; 278 #endif 279 280 #ifdef FASTEST 281 if (level != 0) level = 1; 282 #else 283 if (level == Z_DEFAULT_COMPRESSION) level = 6; 284 #endif 285 286 if (windowBits < 0) { /* suppress zlib wrapper */ 287 wrap = 0; 288 windowBits = -windowBits; 289 } 290 #ifdef GZIP 291 else if (windowBits > 15) { 292 wrap = 2; /* write gzip wrapper instead */ 293 windowBits -= 16; 294 } 295 #endif 296 if (memLevel < 1 || memLevel > MAX_MEM_LEVEL || method != Z_DEFLATED || 297 windowBits < 8 || windowBits > 15 || level < 0 || level > 9 || 298 strategy < 0 || strategy > Z_FIXED || (windowBits == 8 && wrap != 1)) { 299 return Z_STREAM_ERROR; 300 } 301 if (windowBits == 8) windowBits = 9; /* until 256-byte window bug fixed */ 302 s = (deflate_state *) ZALLOC(strm, 1, sizeof(deflate_state)); 303 if (s == Z_NULL) return Z_MEM_ERROR; 304 strm->state = (struct internal_state FAR *)s; 305 s->strm = strm; 306 s->status = INIT_STATE; /* to pass state test in deflateReset() */ 307 308 s->wrap = wrap; 309 s->gzhead = Z_NULL; 310 s->w_bits = (uInt)windowBits; 311 s->w_size = 1 << s->w_bits; 312 s->w_mask = s->w_size - 1; 313 314 s->hash_bits = (uInt)memLevel + 7; 315 s->hash_size = 1 << s->hash_bits; 316 s->hash_mask = s->hash_size - 1; 317 s->hash_shift = ((s->hash_bits+MIN_MATCH-1)/MIN_MATCH); 318 319 s->window = (Bytef *) ZALLOC(strm, s->w_size, 2*sizeof(Byte)); 320 s->prev = (Posf *) ZALLOC(strm, s->w_size, sizeof(Pos)); 321 s->head = (Posf *) ZALLOC(strm, s->hash_size, sizeof(Pos)); 322 323 s->high_water = 0; /* nothing written to s->window yet */ 324 325 s->lit_bufsize = 1 << (memLevel + 6); /* 16K elements by default */ 326 327 /* We overlay pending_buf and sym_buf. This works since the average size 328 * for length/distance pairs over any compressed block is assured to be 31 329 * bits or less. 330 * 331 * Analysis: The longest fixed codes are a length code of 8 bits plus 5 332 * extra bits, for lengths 131 to 257. The longest fixed distance codes are 333 * 5 bits plus 13 extra bits, for distances 16385 to 32768. The longest 334 * possible fixed-codes length/distance pair is then 31 bits total. 335 * 336 * sym_buf starts one-fourth of the way into pending_buf. So there are 337 * three bytes in sym_buf for every four bytes in pending_buf. Each symbol 338 * in sym_buf is three bytes -- two for the distance and one for the 339 * literal/length. As each symbol is consumed, the pointer to the next 340 * sym_buf value to read moves forward three bytes. From that symbol, up to 341 * 31 bits are written to pending_buf. The closest the written pending_buf 342 * bits gets to the next sym_buf symbol to read is just before the last 343 * code is written. At that time, 31*(n-2) bits have been written, just 344 * after 24*(n-2) bits have been consumed from sym_buf. sym_buf starts at 345 * 8*n bits into pending_buf. (Note that the symbol buffer fills when n-1 346 * symbols are written.) The closest the writing gets to what is unread is 347 * then n+14 bits. Here n is lit_bufsize, which is 16384 by default, and 348 * can range from 128 to 32768. 349 * 350 * Therefore, at a minimum, there are 142 bits of space between what is 351 * written and what is read in the overlain buffers, so the symbols cannot 352 * be overwritten by the compressed data. That space is actually 139 bits, 353 * due to the three-bit fixed-code block header. 354 * 355 * That covers the case where either Z_FIXED is specified, forcing fixed 356 * codes, or when the use of fixed codes is chosen, because that choice 357 * results in a smaller compressed block than dynamic codes. That latter 358 * condition then assures that the above analysis also covers all dynamic 359 * blocks. A dynamic-code block will only be chosen to be emitted if it has 360 * fewer bits than a fixed-code block would for the same set of symbols. 361 * Therefore its average symbol length is assured to be less than 31. So 362 * the compressed data for a dynamic block also cannot overwrite the 363 * symbols from which it is being constructed. 364 */ 365 366 s->pending_buf = (uchf *) ZALLOC(strm, s->lit_bufsize, 4); 367 s->pending_buf_size = (ulg)s->lit_bufsize * 4; 368 369 if (s->window == Z_NULL || s->prev == Z_NULL || s->head == Z_NULL || 370 s->pending_buf == Z_NULL) { 371 s->status = FINISH_STATE; 372 strm->msg = ERR_MSG(Z_MEM_ERROR); 373 deflateEnd (strm); 374 return Z_MEM_ERROR; 375 } 376 s->sym_buf = s->pending_buf + s->lit_bufsize; 377 s->sym_end = (s->lit_bufsize - 1) * 3; 378 /* We avoid equality with lit_bufsize*3 because of wraparound at 64K 379 * on 16 bit machines and because stored blocks are restricted to 380 * 64K-1 bytes. 381 */ 382 383 s->level = level; 384 s->strategy = strategy; 385 s->method = (Byte)method; 386 387 return deflateReset(strm); 388 } 389 390 /* ========================================================================= 391 * Check for a valid deflate stream state. Return 0 if ok, 1 if not. 392 */ 393 local int deflateStateCheck (strm) 394 z_streamp strm; 395 { 396 deflate_state *s; 397 if (strm == Z_NULL || 398 strm->zalloc == (alloc_func)0 || strm->zfree == (free_func)0) 399 return 1; 400 s = strm->state; 401 if (s == Z_NULL || s->strm != strm || (s->status != INIT_STATE && 402 #ifdef GZIP 403 s->status != GZIP_STATE && 404 #endif 405 s->status != EXTRA_STATE && 406 s->status != NAME_STATE && 407 s->status != COMMENT_STATE && 408 s->status != HCRC_STATE && 409 s->status != BUSY_STATE && 410 s->status != FINISH_STATE)) 411 return 1; 412 return 0; 413 } 414 415 /* ========================================================================= */ 416 int ZEXPORT deflateSetDictionary (strm, dictionary, dictLength) 417 z_streamp strm; 418 const Bytef *dictionary; 419 uInt dictLength; 420 { 421 deflate_state *s; 422 uInt str, n; 423 int wrap; 424 unsigned avail; 425 z_const unsigned char *next; 426 427 if (deflateStateCheck(strm) || dictionary == Z_NULL) 428 return Z_STREAM_ERROR; 429 s = strm->state; 430 wrap = s->wrap; 431 if (wrap == 2 || (wrap == 1 && s->status != INIT_STATE) || s->lookahead) 432 return Z_STREAM_ERROR; 433 434 /* when using zlib wrappers, compute Adler-32 for provided dictionary */ 435 if (wrap == 1) 436 strm->adler = adler32(strm->adler, dictionary, dictLength); 437 s->wrap = 0; /* avoid computing Adler-32 in read_buf */ 438 439 /* if dictionary would fill window, just replace the history */ 440 if (dictLength >= s->w_size) { 441 if (wrap == 0) { /* already empty otherwise */ 442 CLEAR_HASH(s); 443 s->strstart = 0; 444 s->block_start = 0L; 445 s->insert = 0; 446 } 447 dictionary += dictLength - s->w_size; /* use the tail */ 448 dictLength = s->w_size; 449 } 450 451 /* insert dictionary into window and hash */ 452 avail = strm->avail_in; 453 next = strm->next_in; 454 strm->avail_in = dictLength; 455 strm->next_in = (z_const Bytef *)dictionary; 456 fill_window(s); 457 while (s->lookahead >= MIN_MATCH) { 458 str = s->strstart; 459 n = s->lookahead - (MIN_MATCH-1); 460 do { 461 UPDATE_HASH(s, s->ins_h, s->window[str + MIN_MATCH-1]); 462 #ifndef FASTEST 463 s->prev[str & s->w_mask] = s->head[s->ins_h]; 464 #endif 465 s->head[s->ins_h] = (Pos)str; 466 str++; 467 } while (--n); 468 s->strstart = str; 469 s->lookahead = MIN_MATCH-1; 470 fill_window(s); 471 } 472 s->strstart += s->lookahead; 473 s->block_start = (long)s->strstart; 474 s->insert = s->lookahead; 475 s->lookahead = 0; 476 s->match_length = s->prev_length = MIN_MATCH-1; 477 s->match_available = 0; 478 strm->next_in = next; 479 strm->avail_in = avail; 480 s->wrap = wrap; 481 return Z_OK; 482 } 483 484 /* ========================================================================= */ 485 int ZEXPORT deflateGetDictionary (strm, dictionary, dictLength) 486 z_streamp strm; 487 Bytef *dictionary; 488 uInt *dictLength; 489 { 490 deflate_state *s; 491 uInt len; 492 493 if (deflateStateCheck(strm)) 494 return Z_STREAM_ERROR; 495 s = strm->state; 496 len = s->strstart + s->lookahead; 497 if (len > s->w_size) 498 len = s->w_size; 499 if (dictionary != Z_NULL && len) 500 zmemcpy(dictionary, s->window + s->strstart + s->lookahead - len, len); 501 if (dictLength != Z_NULL) 502 *dictLength = len; 503 return Z_OK; 504 } 505 506 /* ========================================================================= */ 507 int ZEXPORT deflateResetKeep (strm) 508 z_streamp strm; 509 { 510 deflate_state *s; 511 512 if (deflateStateCheck(strm)) { 513 return Z_STREAM_ERROR; 514 } 515 516 strm->total_in = strm->total_out = 0; 517 strm->msg = Z_NULL; /* use zfree if we ever allocate msg dynamically */ 518 strm->data_type = Z_UNKNOWN; 519 520 s = (deflate_state *)strm->state; 521 s->pending = 0; 522 s->pending_out = s->pending_buf; 523 524 if (s->wrap < 0) { 525 s->wrap = -s->wrap; /* was made negative by deflate(..., Z_FINISH); */ 526 } 527 s->status = 528 #ifdef GZIP 529 s->wrap == 2 ? GZIP_STATE : 530 #endif 531 INIT_STATE; 532 strm->adler = 533 #ifdef GZIP 534 s->wrap == 2 ? crc32(0L, Z_NULL, 0) : 535 #endif 536 adler32(0L, Z_NULL, 0); 537 s->last_flush = -2; 538 539 _tr_init(s); 540 541 return Z_OK; 542 } 543 544 /* ========================================================================= */ 545 int ZEXPORT deflateReset (strm) 546 z_streamp strm; 547 { 548 int ret; 549 550 ret = deflateResetKeep(strm); 551 if (ret == Z_OK) 552 lm_init(strm->state); 553 return ret; 554 } 555 556 /* ========================================================================= */ 557 int ZEXPORT deflateSetHeader (strm, head) 558 z_streamp strm; 559 gz_headerp head; 560 { 561 if (deflateStateCheck(strm) || strm->state->wrap != 2) 562 return Z_STREAM_ERROR; 563 strm->state->gzhead = head; 564 return Z_OK; 565 } 566 567 /* ========================================================================= */ 568 int ZEXPORT deflatePending (strm, pending, bits) 569 unsigned *pending; 570 int *bits; 571 z_streamp strm; 572 { 573 if (deflateStateCheck(strm)) return Z_STREAM_ERROR; 574 if (pending != Z_NULL) 575 *pending = strm->state->pending; 576 if (bits != Z_NULL) 577 *bits = strm->state->bi_valid; 578 return Z_OK; 579 } 580 581 /* ========================================================================= */ 582 int ZEXPORT deflatePrime (strm, bits, value) 583 z_streamp strm; 584 int bits; 585 int value; 586 { 587 deflate_state *s; 588 int put; 589 590 if (deflateStateCheck(strm)) return Z_STREAM_ERROR; 591 s = strm->state; 592 if (bits < 0 || bits > 16 || 593 s->sym_buf < s->pending_out + ((Buf_size + 7) >> 3)) 594 return Z_BUF_ERROR; 595 do { 596 put = Buf_size - s->bi_valid; 597 if (put > bits) 598 put = bits; 599 s->bi_buf |= (ush)((value & ((1 << put) - 1)) << s->bi_valid); 600 s->bi_valid += put; 601 _tr_flush_bits(s); 602 value >>= put; 603 bits -= put; 604 } while (bits); 605 return Z_OK; 606 } 607 608 /* ========================================================================= */ 609 int ZEXPORT deflateParams(strm, level, strategy) 610 z_streamp strm; 611 int level; 612 int strategy; 613 { 614 deflate_state *s; 615 compress_func func; 616 617 if (deflateStateCheck(strm)) return Z_STREAM_ERROR; 618 s = strm->state; 619 620 #ifdef FASTEST 621 if (level != 0) level = 1; 622 #else 623 if (level == Z_DEFAULT_COMPRESSION) level = 6; 624 #endif 625 if (level < 0 || level > 9 || strategy < 0 || strategy > Z_FIXED) { 626 return Z_STREAM_ERROR; 627 } 628 func = configuration_table[s->level].func; 629 630 if ((strategy != s->strategy || func != configuration_table[level].func) && 631 s->last_flush != -2) { 632 /* Flush the last buffer: */ 633 int err = deflate(strm, Z_BLOCK); 634 if (err == Z_STREAM_ERROR) 635 return err; 636 if (strm->avail_in || (s->strstart - s->block_start) + s->lookahead) 637 return Z_BUF_ERROR; 638 } 639 if (s->level != level) { 640 if (s->level == 0 && s->matches != 0) { 641 if (s->matches == 1) 642 slide_hash(s); 643 else 644 CLEAR_HASH(s); 645 s->matches = 0; 646 } 647 s->level = level; 648 s->max_lazy_match = configuration_table[level].max_lazy; 649 s->good_match = configuration_table[level].good_length; 650 s->nice_match = configuration_table[level].nice_length; 651 s->max_chain_length = configuration_table[level].max_chain; 652 } 653 s->strategy = strategy; 654 return Z_OK; 655 } 656 657 /* ========================================================================= */ 658 int ZEXPORT deflateTune(strm, good_length, max_lazy, nice_length, max_chain) 659 z_streamp strm; 660 int good_length; 661 int max_lazy; 662 int nice_length; 663 int max_chain; 664 { 665 deflate_state *s; 666 667 if (deflateStateCheck(strm)) return Z_STREAM_ERROR; 668 s = strm->state; 669 s->good_match = (uInt)good_length; 670 s->max_lazy_match = (uInt)max_lazy; 671 s->nice_match = nice_length; 672 s->max_chain_length = (uInt)max_chain; 673 return Z_OK; 674 } 675 676 /* ========================================================================= 677 * For the default windowBits of 15 and memLevel of 8, this function returns 678 * a close to exact, as well as small, upper bound on the compressed size. 679 * They are coded as constants here for a reason--if the #define's are 680 * changed, then this function needs to be changed as well. The return 681 * value for 15 and 8 only works for those exact settings. 682 * 683 * For any setting other than those defaults for windowBits and memLevel, 684 * the value returned is a conservative worst case for the maximum expansion 685 * resulting from using fixed blocks instead of stored blocks, which deflate 686 * can emit on compressed data for some combinations of the parameters. 687 * 688 * This function could be more sophisticated to provide closer upper bounds for 689 * every combination of windowBits and memLevel. But even the conservative 690 * upper bound of about 14% expansion does not seem onerous for output buffer 691 * allocation. 692 */ 693 uLong ZEXPORT deflateBound(strm, sourceLen) 694 z_streamp strm; 695 uLong sourceLen; 696 { 697 deflate_state *s; 698 uLong complen, wraplen; 699 700 /* conservative upper bound for compressed data */ 701 complen = sourceLen + 702 ((sourceLen + 7) >> 3) + ((sourceLen + 63) >> 6) + 5; 703 704 /* if can't get parameters, return conservative bound plus zlib wrapper */ 705 if (deflateStateCheck(strm)) 706 return complen + 6; 707 708 /* compute wrapper length */ 709 s = strm->state; 710 switch (s->wrap) { 711 case 0: /* raw deflate */ 712 wraplen = 0; 713 break; 714 case 1: /* zlib wrapper */ 715 wraplen = 6 + (s->strstart ? 4 : 0); 716 break; 717 #ifdef GZIP 718 case 2: /* gzip wrapper */ 719 wraplen = 18; 720 if (s->gzhead != Z_NULL) { /* user-supplied gzip header */ 721 Bytef *str; 722 if (s->gzhead->extra != Z_NULL) 723 wraplen += 2 + s->gzhead->extra_len; 724 str = s->gzhead->name; 725 if (str != Z_NULL) 726 do { 727 wraplen++; 728 } while (*str++); 729 str = s->gzhead->comment; 730 if (str != Z_NULL) 731 do { 732 wraplen++; 733 } while (*str++); 734 if (s->gzhead->hcrc) 735 wraplen += 2; 736 } 737 break; 738 #endif 739 default: /* for compiler happiness */ 740 wraplen = 6; 741 } 742 743 /* if not default parameters, return conservative bound */ 744 if (s->w_bits != 15 || s->hash_bits != 8 + 7) 745 return complen + wraplen; 746 747 /* default settings: return tight bound for that case */ 748 return sourceLen + (sourceLen >> 12) + (sourceLen >> 14) + 749 (sourceLen >> 25) + 13 - 6 + wraplen; 750 } 751 752 /* ========================================================================= 753 * Put a short in the pending buffer. The 16-bit value is put in MSB order. 754 * IN assertion: the stream state is correct and there is enough room in 755 * pending_buf. 756 */ 757 local void putShortMSB (s, b) 758 deflate_state *s; 759 uInt b; 760 { 761 put_byte(s, (Byte)(b >> 8)); 762 put_byte(s, (Byte)(b & 0xff)); 763 } 764 765 /* ========================================================================= 766 * Flush as much pending output as possible. All deflate() output, except for 767 * some deflate_stored() output, goes through this function so some 768 * applications may wish to modify it to avoid allocating a large 769 * strm->next_out buffer and copying into it. (See also read_buf()). 770 */ 771 local void flush_pending(strm) 772 z_streamp strm; 773 { 774 unsigned len; 775 deflate_state *s = strm->state; 776 777 _tr_flush_bits(s); 778 len = s->pending; 779 if (len > strm->avail_out) len = strm->avail_out; 780 if (len == 0) return; 781 782 zmemcpy(strm->next_out, s->pending_out, len); 783 strm->next_out += len; 784 s->pending_out += len; 785 strm->total_out += len; 786 strm->avail_out -= len; 787 s->pending -= len; 788 if (s->pending == 0) { 789 s->pending_out = s->pending_buf; 790 } 791 } 792 793 /* =========================================================================== 794 * Update the header CRC with the bytes s->pending_buf[beg..s->pending - 1]. 795 */ 796 #define HCRC_UPDATE(beg) \ 797 do { \ 798 if (s->gzhead->hcrc && s->pending > (beg)) \ 799 strm->adler = crc32(strm->adler, s->pending_buf + (beg), \ 800 s->pending - (beg)); \ 801 } while (0) 802 803 /* ========================================================================= */ 804 int ZEXPORT deflate (strm, flush) 805 z_streamp strm; 806 int flush; 807 { 808 int old_flush; /* value of flush param for previous deflate call */ 809 deflate_state *s; 810 811 if (deflateStateCheck(strm) || flush > Z_BLOCK || flush < 0) { 812 return Z_STREAM_ERROR; 813 } 814 s = strm->state; 815 816 if (strm->next_out == Z_NULL || 817 (strm->avail_in != 0 && strm->next_in == Z_NULL) || 818 (s->status == FINISH_STATE && flush != Z_FINISH)) { 819 ERR_RETURN(strm, Z_STREAM_ERROR); 820 } 821 if (strm->avail_out == 0) ERR_RETURN(strm, Z_BUF_ERROR); 822 823 old_flush = s->last_flush; 824 s->last_flush = flush; 825 826 /* Flush as much pending output as possible */ 827 if (s->pending != 0) { 828 flush_pending(strm); 829 if (strm->avail_out == 0) { 830 /* Since avail_out is 0, deflate will be called again with 831 * more output space, but possibly with both pending and 832 * avail_in equal to zero. There won't be anything to do, 833 * but this is not an error situation so make sure we 834 * return OK instead of BUF_ERROR at next call of deflate: 835 */ 836 s->last_flush = -1; 837 return Z_OK; 838 } 839 840 /* Make sure there is something to do and avoid duplicate consecutive 841 * flushes. For repeated and useless calls with Z_FINISH, we keep 842 * returning Z_STREAM_END instead of Z_BUF_ERROR. 843 */ 844 } else if (strm->avail_in == 0 && RANK(flush) <= RANK(old_flush) && 845 flush != Z_FINISH) { 846 ERR_RETURN(strm, Z_BUF_ERROR); 847 } 848 849 /* User must not provide more input after the first FINISH: */ 850 if (s->status == FINISH_STATE && strm->avail_in != 0) { 851 ERR_RETURN(strm, Z_BUF_ERROR); 852 } 853 854 /* Write the header */ 855 if (s->status == INIT_STATE && s->wrap == 0) 856 s->status = BUSY_STATE; 857 if (s->status == INIT_STATE) { 858 /* zlib header */ 859 uInt header = (Z_DEFLATED + ((s->w_bits-8)<<4)) << 8; 860 uInt level_flags; 861 862 if (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2) 863 level_flags = 0; 864 else if (s->level < 6) 865 level_flags = 1; 866 else if (s->level == 6) 867 level_flags = 2; 868 else 869 level_flags = 3; 870 header |= (level_flags << 6); 871 if (s->strstart != 0) header |= PRESET_DICT; 872 header += 31 - (header % 31); 873 874 putShortMSB(s, header); 875 876 /* Save the adler32 of the preset dictionary: */ 877 if (s->strstart != 0) { 878 putShortMSB(s, (uInt)(strm->adler >> 16)); 879 putShortMSB(s, (uInt)(strm->adler & 0xffff)); 880 } 881 strm->adler = adler32(0L, Z_NULL, 0); 882 s->status = BUSY_STATE; 883 884 /* Compression must start with an empty pending buffer */ 885 flush_pending(strm); 886 if (s->pending != 0) { 887 s->last_flush = -1; 888 return Z_OK; 889 } 890 } 891 #ifdef GZIP 892 if (s->status == GZIP_STATE) { 893 /* gzip header */ 894 strm->adler = crc32(0L, Z_NULL, 0); 895 put_byte(s, 31); 896 put_byte(s, 139); 897 put_byte(s, 8); 898 if (s->gzhead == Z_NULL) { 899 put_byte(s, 0); 900 put_byte(s, 0); 901 put_byte(s, 0); 902 put_byte(s, 0); 903 put_byte(s, 0); 904 put_byte(s, s->level == 9 ? 2 : 905 (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ? 906 4 : 0)); 907 put_byte(s, OS_CODE); 908 s->status = BUSY_STATE; 909 910 /* Compression must start with an empty pending buffer */ 911 flush_pending(strm); 912 if (s->pending != 0) { 913 s->last_flush = -1; 914 return Z_OK; 915 } 916 } 917 else { 918 put_byte(s, (s->gzhead->text ? 1 : 0) + 919 (s->gzhead->hcrc ? 2 : 0) + 920 (s->gzhead->extra == Z_NULL ? 0 : 4) + 921 (s->gzhead->name == Z_NULL ? 0 : 8) + 922 (s->gzhead->comment == Z_NULL ? 0 : 16) 923 ); 924 put_byte(s, (Byte)(s->gzhead->time & 0xff)); 925 put_byte(s, (Byte)((s->gzhead->time >> 8) & 0xff)); 926 put_byte(s, (Byte)((s->gzhead->time >> 16) & 0xff)); 927 put_byte(s, (Byte)((s->gzhead->time >> 24) & 0xff)); 928 put_byte(s, s->level == 9 ? 2 : 929 (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ? 930 4 : 0)); 931 put_byte(s, s->gzhead->os & 0xff); 932 if (s->gzhead->extra != Z_NULL) { 933 put_byte(s, s->gzhead->extra_len & 0xff); 934 put_byte(s, (s->gzhead->extra_len >> 8) & 0xff); 935 } 936 if (s->gzhead->hcrc) 937 strm->adler = crc32(strm->adler, s->pending_buf, 938 s->pending); 939 s->gzindex = 0; 940 s->status = EXTRA_STATE; 941 } 942 } 943 if (s->status == EXTRA_STATE) { 944 if (s->gzhead->extra != Z_NULL) { 945 ulg beg = s->pending; /* start of bytes to update crc */ 946 uInt left = (s->gzhead->extra_len & 0xffff) - s->gzindex; 947 while (s->pending + left > s->pending_buf_size) { 948 uInt copy = s->pending_buf_size - s->pending; 949 zmemcpy(s->pending_buf + s->pending, 950 s->gzhead->extra + s->gzindex, copy); 951 s->pending = s->pending_buf_size; 952 HCRC_UPDATE(beg); 953 s->gzindex += copy; 954 flush_pending(strm); 955 if (s->pending != 0) { 956 s->last_flush = -1; 957 return Z_OK; 958 } 959 beg = 0; 960 left -= copy; 961 } 962 zmemcpy(s->pending_buf + s->pending, 963 s->gzhead->extra + s->gzindex, left); 964 s->pending += left; 965 HCRC_UPDATE(beg); 966 s->gzindex = 0; 967 } 968 s->status = NAME_STATE; 969 } 970 if (s->status == NAME_STATE) { 971 if (s->gzhead->name != Z_NULL) { 972 ulg beg = s->pending; /* start of bytes to update crc */ 973 int val; 974 do { 975 if (s->pending == s->pending_buf_size) { 976 HCRC_UPDATE(beg); 977 flush_pending(strm); 978 if (s->pending != 0) { 979 s->last_flush = -1; 980 return Z_OK; 981 } 982 beg = 0; 983 } 984 val = s->gzhead->name[s->gzindex++]; 985 put_byte(s, val); 986 } while (val != 0); 987 HCRC_UPDATE(beg); 988 s->gzindex = 0; 989 } 990 s->status = COMMENT_STATE; 991 } 992 if (s->status == COMMENT_STATE) { 993 if (s->gzhead->comment != Z_NULL) { 994 ulg beg = s->pending; /* start of bytes to update crc */ 995 int val; 996 do { 997 if (s->pending == s->pending_buf_size) { 998 HCRC_UPDATE(beg); 999 flush_pending(strm); 1000 if (s->pending != 0) { 1001 s->last_flush = -1; 1002 return Z_OK; 1003 } 1004 beg = 0; 1005 } 1006 val = s->gzhead->comment[s->gzindex++]; 1007 put_byte(s, val); 1008 } while (val != 0); 1009 HCRC_UPDATE(beg); 1010 } 1011 s->status = HCRC_STATE; 1012 } 1013 if (s->status == HCRC_STATE) { 1014 if (s->gzhead->hcrc) { 1015 if (s->pending + 2 > s->pending_buf_size) { 1016 flush_pending(strm); 1017 if (s->pending != 0) { 1018 s->last_flush = -1; 1019 return Z_OK; 1020 } 1021 } 1022 put_byte(s, (Byte)(strm->adler & 0xff)); 1023 put_byte(s, (Byte)((strm->adler >> 8) & 0xff)); 1024 strm->adler = crc32(0L, Z_NULL, 0); 1025 } 1026 s->status = BUSY_STATE; 1027 1028 /* Compression must start with an empty pending buffer */ 1029 flush_pending(strm); 1030 if (s->pending != 0) { 1031 s->last_flush = -1; 1032 return Z_OK; 1033 } 1034 } 1035 #endif 1036 1037 /* Start a new block or continue the current one. 1038 */ 1039 if (strm->avail_in != 0 || s->lookahead != 0 || 1040 (flush != Z_NO_FLUSH && s->status != FINISH_STATE)) { 1041 block_state bstate; 1042 1043 bstate = s->level == 0 ? deflate_stored(s, flush) : 1044 s->strategy == Z_HUFFMAN_ONLY ? deflate_huff(s, flush) : 1045 s->strategy == Z_RLE ? deflate_rle(s, flush) : 1046 (*(configuration_table[s->level].func))(s, flush); 1047 1048 if (bstate == finish_started || bstate == finish_done) { 1049 s->status = FINISH_STATE; 1050 } 1051 if (bstate == need_more || bstate == finish_started) { 1052 if (strm->avail_out == 0) { 1053 s->last_flush = -1; /* avoid BUF_ERROR next call, see above */ 1054 } 1055 return Z_OK; 1056 /* If flush != Z_NO_FLUSH && avail_out == 0, the next call 1057 * of deflate should use the same flush parameter to make sure 1058 * that the flush is complete. So we don't have to output an 1059 * empty block here, this will be done at next call. This also 1060 * ensures that for a very small output buffer, we emit at most 1061 * one empty block. 1062 */ 1063 } 1064 if (bstate == block_done) { 1065 if (flush == Z_PARTIAL_FLUSH) { 1066 _tr_align(s); 1067 } else if (flush != Z_BLOCK) { /* FULL_FLUSH or SYNC_FLUSH */ 1068 _tr_stored_block(s, (char*)0, 0L, 0); 1069 /* For a full flush, this empty block will be recognized 1070 * as a special marker by inflate_sync(). 1071 */ 1072 if (flush == Z_FULL_FLUSH) { 1073 CLEAR_HASH(s); /* forget history */ 1074 if (s->lookahead == 0) { 1075 s->strstart = 0; 1076 s->block_start = 0L; 1077 s->insert = 0; 1078 } 1079 } 1080 } 1081 flush_pending(strm); 1082 if (strm->avail_out == 0) { 1083 s->last_flush = -1; /* avoid BUF_ERROR at next call, see above */ 1084 return Z_OK; 1085 } 1086 } 1087 } 1088 1089 if (flush != Z_FINISH) return Z_OK; 1090 if (s->wrap <= 0) return Z_STREAM_END; 1091 1092 /* Write the trailer */ 1093 #ifdef GZIP 1094 if (s->wrap == 2) { 1095 put_byte(s, (Byte)(strm->adler & 0xff)); 1096 put_byte(s, (Byte)((strm->adler >> 8) & 0xff)); 1097 put_byte(s, (Byte)((strm->adler >> 16) & 0xff)); 1098 put_byte(s, (Byte)((strm->adler >> 24) & 0xff)); 1099 put_byte(s, (Byte)(strm->total_in & 0xff)); 1100 put_byte(s, (Byte)((strm->total_in >> 8) & 0xff)); 1101 put_byte(s, (Byte)((strm->total_in >> 16) & 0xff)); 1102 put_byte(s, (Byte)((strm->total_in >> 24) & 0xff)); 1103 } 1104 else 1105 #endif 1106 { 1107 putShortMSB(s, (uInt)(strm->adler >> 16)); 1108 putShortMSB(s, (uInt)(strm->adler & 0xffff)); 1109 } 1110 flush_pending(strm); 1111 /* If avail_out is zero, the application will call deflate again 1112 * to flush the rest. 1113 */ 1114 if (s->wrap > 0) s->wrap = -s->wrap; /* write the trailer only once! */ 1115 return s->pending != 0 ? Z_OK : Z_STREAM_END; 1116 } 1117 1118 /* ========================================================================= */ 1119 int ZEXPORT deflateEnd (strm) 1120 z_streamp strm; 1121 { 1122 int status; 1123 1124 if (deflateStateCheck(strm)) return Z_STREAM_ERROR; 1125 1126 status = strm->state->status; 1127 1128 /* Deallocate in reverse order of allocations: */ 1129 TRY_FREE(strm, strm->state->pending_buf); 1130 TRY_FREE(strm, strm->state->head); 1131 TRY_FREE(strm, strm->state->prev); 1132 TRY_FREE(strm, strm->state->window); 1133 1134 ZFREE(strm, strm->state); 1135 strm->state = Z_NULL; 1136 1137 return status == BUSY_STATE ? Z_DATA_ERROR : Z_OK; 1138 } 1139 1140 /* ========================================================================= 1141 * Copy the source state to the destination state. 1142 * To simplify the source, this is not supported for 16-bit MSDOS (which 1143 * doesn't have enough memory anyway to duplicate compression states). 1144 */ 1145 int ZEXPORT deflateCopy (dest, source) 1146 z_streamp dest; 1147 z_streamp source; 1148 { 1149 #ifdef MAXSEG_64K 1150 return Z_STREAM_ERROR; 1151 #else 1152 deflate_state *ds; 1153 deflate_state *ss; 1154 1155 1156 if (deflateStateCheck(source) || dest == Z_NULL) { 1157 return Z_STREAM_ERROR; 1158 } 1159 1160 ss = source->state; 1161 1162 zmemcpy((voidpf)dest, (voidpf)source, sizeof(z_stream)); 1163 1164 ds = (deflate_state *) ZALLOC(dest, 1, sizeof(deflate_state)); 1165 if (ds == Z_NULL) return Z_MEM_ERROR; 1166 dest->state = (struct internal_state FAR *) ds; 1167 zmemcpy((voidpf)ds, (voidpf)ss, sizeof(deflate_state)); 1168 ds->strm = dest; 1169 1170 ds->window = (Bytef *) ZALLOC(dest, ds->w_size, 2*sizeof(Byte)); 1171 ds->prev = (Posf *) ZALLOC(dest, ds->w_size, sizeof(Pos)); 1172 ds->head = (Posf *) ZALLOC(dest, ds->hash_size, sizeof(Pos)); 1173 ds->pending_buf = (uchf *) ZALLOC(dest, ds->lit_bufsize, 4); 1174 1175 if (ds->window == Z_NULL || ds->prev == Z_NULL || ds->head == Z_NULL || 1176 ds->pending_buf == Z_NULL) { 1177 deflateEnd (dest); 1178 return Z_MEM_ERROR; 1179 } 1180 /* following zmemcpy do not work for 16-bit MSDOS */ 1181 zmemcpy(ds->window, ss->window, ds->w_size * 2 * sizeof(Byte)); 1182 zmemcpy((voidpf)ds->prev, (voidpf)ss->prev, ds->w_size * sizeof(Pos)); 1183 zmemcpy((voidpf)ds->head, (voidpf)ss->head, ds->hash_size * sizeof(Pos)); 1184 zmemcpy(ds->pending_buf, ss->pending_buf, (uInt)ds->pending_buf_size); 1185 1186 ds->pending_out = ds->pending_buf + (ss->pending_out - ss->pending_buf); 1187 ds->sym_buf = ds->pending_buf + ds->lit_bufsize; 1188 1189 ds->l_desc.dyn_tree = ds->dyn_ltree; 1190 ds->d_desc.dyn_tree = ds->dyn_dtree; 1191 ds->bl_desc.dyn_tree = ds->bl_tree; 1192 1193 return Z_OK; 1194 #endif /* MAXSEG_64K */ 1195 } 1196 1197 /* =========================================================================== 1198 * Read a new buffer from the current input stream, update the adler32 1199 * and total number of bytes read. All deflate() input goes through 1200 * this function so some applications may wish to modify it to avoid 1201 * allocating a large strm->next_in buffer and copying from it. 1202 * (See also flush_pending()). 1203 */ 1204 local unsigned read_buf(strm, buf, size) 1205 z_streamp strm; 1206 Bytef *buf; 1207 unsigned size; 1208 { 1209 unsigned len = strm->avail_in; 1210 1211 if (len > size) len = size; 1212 if (len == 0) return 0; 1213 1214 strm->avail_in -= len; 1215 1216 zmemcpy(buf, strm->next_in, len); 1217 if (strm->state->wrap == 1) { 1218 strm->adler = adler32(strm->adler, buf, len); 1219 } 1220 #ifdef GZIP 1221 else if (strm->state->wrap == 2) { 1222 strm->adler = crc32(strm->adler, buf, len); 1223 } 1224 #endif 1225 strm->next_in += len; 1226 strm->total_in += len; 1227 1228 return len; 1229 } 1230 1231 /* =========================================================================== 1232 * Initialize the "longest match" routines for a new zlib stream 1233 */ 1234 local void lm_init (s) 1235 deflate_state *s; 1236 { 1237 s->window_size = (ulg)2L*s->w_size; 1238 1239 CLEAR_HASH(s); 1240 1241 /* Set the default configuration parameters: 1242 */ 1243 s->max_lazy_match = configuration_table[s->level].max_lazy; 1244 s->good_match = configuration_table[s->level].good_length; 1245 s->nice_match = configuration_table[s->level].nice_length; 1246 s->max_chain_length = configuration_table[s->level].max_chain; 1247 1248 s->strstart = 0; 1249 s->block_start = 0L; 1250 s->lookahead = 0; 1251 s->insert = 0; 1252 s->match_length = s->prev_length = MIN_MATCH-1; 1253 s->match_available = 0; 1254 s->ins_h = 0; 1255 #ifndef FASTEST 1256 #ifdef ASMV 1257 match_init(); /* initialize the asm code */ 1258 #endif 1259 #endif 1260 } 1261 1262 #ifndef FASTEST 1263 /* =========================================================================== 1264 * Set match_start to the longest match starting at the given string and 1265 * return its length. Matches shorter or equal to prev_length are discarded, 1266 * in which case the result is equal to prev_length and match_start is 1267 * garbage. 1268 * IN assertions: cur_match is the head of the hash chain for the current 1269 * string (strstart) and its distance is <= MAX_DIST, and prev_length >= 1 1270 * OUT assertion: the match length is not greater than s->lookahead. 1271 */ 1272 #ifndef ASMV 1273 /* For 80x86 and 680x0, an optimized version will be provided in match.asm or 1274 * match.S. The code will be functionally equivalent. 1275 */ 1276 local uInt longest_match(s, cur_match) 1277 deflate_state *s; 1278 IPos cur_match; /* current match */ 1279 { 1280 unsigned chain_length = s->max_chain_length;/* max hash chain length */ 1281 register Bytef *scan = s->window + s->strstart; /* current string */ 1282 register Bytef *match; /* matched string */ 1283 register int len; /* length of current match */ 1284 int best_len = (int)s->prev_length; /* best match length so far */ 1285 int nice_match = s->nice_match; /* stop if match long enough */ 1286 IPos limit = s->strstart > (IPos)MAX_DIST(s) ? 1287 s->strstart - (IPos)MAX_DIST(s) : NIL; 1288 /* Stop when cur_match becomes <= limit. To simplify the code, 1289 * we prevent matches with the string of window index 0. 1290 */ 1291 Posf *prev = s->prev; 1292 uInt wmask = s->w_mask; 1293 1294 #ifdef UNALIGNED_OK 1295 /* Compare two bytes at a time. Note: this is not always beneficial. 1296 * Try with and without -DUNALIGNED_OK to check. 1297 */ 1298 register Bytef *strend = s->window + s->strstart + MAX_MATCH - 1; 1299 register ush scan_start = *(ushf*)scan; 1300 register ush scan_end = *(ushf*)(scan+best_len-1); 1301 #else 1302 register Bytef *strend = s->window + s->strstart + MAX_MATCH; 1303 register Byte scan_end1 = scan[best_len-1]; 1304 register Byte scan_end = scan[best_len]; 1305 #endif 1306 1307 /* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16. 1308 * It is easy to get rid of this optimization if necessary. 1309 */ 1310 Assert(s->hash_bits >= 8 && MAX_MATCH == 258, "Code too clever"); 1311 1312 /* Do not waste too much time if we already have a good match: */ 1313 if (s->prev_length >= s->good_match) { 1314 chain_length >>= 2; 1315 } 1316 /* Do not look for matches beyond the end of the input. This is necessary 1317 * to make deflate deterministic. 1318 */ 1319 if ((uInt)nice_match > s->lookahead) nice_match = (int)s->lookahead; 1320 1321 Assert((ulg)s->strstart <= s->window_size-MIN_LOOKAHEAD, "need lookahead"); 1322 1323 do { 1324 Assert(cur_match < s->strstart, "no future"); 1325 match = s->window + cur_match; 1326 1327 /* Skip to next match if the match length cannot increase 1328 * or if the match length is less than 2. Note that the checks below 1329 * for insufficient lookahead only occur occasionally for performance 1330 * reasons. Therefore uninitialized memory will be accessed, and 1331 * conditional jumps will be made that depend on those values. 1332 * However the length of the match is limited to the lookahead, so 1333 * the output of deflate is not affected by the uninitialized values. 1334 */ 1335 #if (defined(UNALIGNED_OK) && MAX_MATCH == 258) 1336 /* This code assumes sizeof(unsigned short) == 2. Do not use 1337 * UNALIGNED_OK if your compiler uses a different size. 1338 */ 1339 if (*(ushf*)(match+best_len-1) != scan_end || 1340 *(ushf*)match != scan_start) continue; 1341 1342 /* It is not necessary to compare scan[2] and match[2] since they are 1343 * always equal when the other bytes match, given that the hash keys 1344 * are equal and that HASH_BITS >= 8. Compare 2 bytes at a time at 1345 * strstart+3, +5, ... up to strstart+257. We check for insufficient 1346 * lookahead only every 4th comparison; the 128th check will be made 1347 * at strstart+257. If MAX_MATCH-2 is not a multiple of 8, it is 1348 * necessary to put more guard bytes at the end of the window, or 1349 * to check more often for insufficient lookahead. 1350 */ 1351 Assert(scan[2] == match[2], "scan[2]?"); 1352 scan++, match++; 1353 do { 1354 } while (*(ushf*)(scan+=2) == *(ushf*)(match+=2) && 1355 *(ushf*)(scan+=2) == *(ushf*)(match+=2) && 1356 *(ushf*)(scan+=2) == *(ushf*)(match+=2) && 1357 *(ushf*)(scan+=2) == *(ushf*)(match+=2) && 1358 scan < strend); 1359 /* The funny "do {}" generates better code on most compilers */ 1360 1361 /* Here, scan <= window+strstart+257 */ 1362 Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan"); 1363 if (*scan == *match) scan++; 1364 1365 len = (MAX_MATCH - 1) - (int)(strend-scan); 1366 scan = strend - (MAX_MATCH-1); 1367 1368 #else /* UNALIGNED_OK */ 1369 1370 if (match[best_len] != scan_end || 1371 match[best_len-1] != scan_end1 || 1372 *match != *scan || 1373 *++match != scan[1]) continue; 1374 1375 /* The check at best_len-1 can be removed because it will be made 1376 * again later. (This heuristic is not always a win.) 1377 * It is not necessary to compare scan[2] and match[2] since they 1378 * are always equal when the other bytes match, given that 1379 * the hash keys are equal and that HASH_BITS >= 8. 1380 */ 1381 scan += 2, match++; 1382 Assert(*scan == *match, "match[2]?"); 1383 1384 /* We check for insufficient lookahead only every 8th comparison; 1385 * the 256th check will be made at strstart+258. 1386 */ 1387 do { 1388 } while (*++scan == *++match && *++scan == *++match && 1389 *++scan == *++match && *++scan == *++match && 1390 *++scan == *++match && *++scan == *++match && 1391 *++scan == *++match && *++scan == *++match && 1392 scan < strend); 1393 1394 Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan"); 1395 1396 len = MAX_MATCH - (int)(strend - scan); 1397 scan = strend - MAX_MATCH; 1398 1399 #endif /* UNALIGNED_OK */ 1400 1401 if (len > best_len) { 1402 s->match_start = cur_match; 1403 best_len = len; 1404 if (len >= nice_match) break; 1405 #ifdef UNALIGNED_OK 1406 scan_end = *(ushf*)(scan+best_len-1); 1407 #else 1408 scan_end1 = scan[best_len-1]; 1409 scan_end = scan[best_len]; 1410 #endif 1411 } 1412 } while ((cur_match = prev[cur_match & wmask]) > limit 1413 && --chain_length != 0); 1414 1415 if ((uInt)best_len <= s->lookahead) return (uInt)best_len; 1416 return s->lookahead; 1417 } 1418 #endif /* ASMV */ 1419 1420 #else /* FASTEST */ 1421 1422 /* --------------------------------------------------------------------------- 1423 * Optimized version for FASTEST only 1424 */ 1425 local uInt longest_match(s, cur_match) 1426 deflate_state *s; 1427 IPos cur_match; /* current match */ 1428 { 1429 register Bytef *scan = s->window + s->strstart; /* current string */ 1430 register Bytef *match; /* matched string */ 1431 register int len; /* length of current match */ 1432 register Bytef *strend = s->window + s->strstart + MAX_MATCH; 1433 1434 /* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16. 1435 * It is easy to get rid of this optimization if necessary. 1436 */ 1437 Assert(s->hash_bits >= 8 && MAX_MATCH == 258, "Code too clever"); 1438 1439 Assert((ulg)s->strstart <= s->window_size-MIN_LOOKAHEAD, "need lookahead"); 1440 1441 Assert(cur_match < s->strstart, "no future"); 1442 1443 match = s->window + cur_match; 1444 1445 /* Return failure if the match length is less than 2: 1446 */ 1447 if (match[0] != scan[0] || match[1] != scan[1]) return MIN_MATCH-1; 1448 1449 /* The check at best_len-1 can be removed because it will be made 1450 * again later. (This heuristic is not always a win.) 1451 * It is not necessary to compare scan[2] and match[2] since they 1452 * are always equal when the other bytes match, given that 1453 * the hash keys are equal and that HASH_BITS >= 8. 1454 */ 1455 scan += 2, match += 2; 1456 Assert(*scan == *match, "match[2]?"); 1457 1458 /* We check for insufficient lookahead only every 8th comparison; 1459 * the 256th check will be made at strstart+258. 1460 */ 1461 do { 1462 } while (*++scan == *++match && *++scan == *++match && 1463 *++scan == *++match && *++scan == *++match && 1464 *++scan == *++match && *++scan == *++match && 1465 *++scan == *++match && *++scan == *++match && 1466 scan < strend); 1467 1468 Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan"); 1469 1470 len = MAX_MATCH - (int)(strend - scan); 1471 1472 if (len < MIN_MATCH) return MIN_MATCH - 1; 1473 1474 s->match_start = cur_match; 1475 return (uInt)len <= s->lookahead ? (uInt)len : s->lookahead; 1476 } 1477 1478 #endif /* FASTEST */ 1479 1480 #ifdef ZLIB_DEBUG 1481 1482 #define EQUAL 0 1483 /* result of memcmp for equal strings */ 1484 1485 /* =========================================================================== 1486 * Check that the match at match_start is indeed a match. 1487 */ 1488 local void check_match(s, start, match, length) 1489 deflate_state *s; 1490 IPos start, match; 1491 int length; 1492 { 1493 /* check that the match is indeed a match */ 1494 if (zmemcmp(s->window + match, 1495 s->window + start, length) != EQUAL) { 1496 fprintf(stderr, " start %u, match %u, length %d\n", 1497 start, match, length); 1498 do { 1499 fprintf(stderr, "%c%c", s->window[match++], s->window[start++]); 1500 } while (--length != 0); 1501 z_error("invalid match"); 1502 } 1503 if (z_verbose > 1) { 1504 fprintf(stderr,"\\[%d,%d]", start-match, length); 1505 do { putc(s->window[start++], stderr); } while (--length != 0); 1506 } 1507 } 1508 #else 1509 # define check_match(s, start, match, length) 1510 #endif /* ZLIB_DEBUG */ 1511 1512 /* =========================================================================== 1513 * Fill the window when the lookahead becomes insufficient. 1514 * Updates strstart and lookahead. 1515 * 1516 * IN assertion: lookahead < MIN_LOOKAHEAD 1517 * OUT assertions: strstart <= window_size-MIN_LOOKAHEAD 1518 * At least one byte has been read, or avail_in == 0; reads are 1519 * performed for at least two bytes (required for the zip translate_eol 1520 * option -- not supported here). 1521 */ 1522 local void fill_window(s) 1523 deflate_state *s; 1524 { 1525 unsigned n; 1526 unsigned more; /* Amount of free space at the end of the window. */ 1527 uInt wsize = s->w_size; 1528 1529 Assert(s->lookahead < MIN_LOOKAHEAD, "already enough lookahead"); 1530 1531 do { 1532 more = (unsigned)(s->window_size -(ulg)s->lookahead -(ulg)s->strstart); 1533 1534 /* Deal with !@#$% 64K limit: */ 1535 if (sizeof(int) <= 2) { 1536 if (more == 0 && s->strstart == 0 && s->lookahead == 0) { 1537 more = wsize; 1538 1539 } else if (more == (unsigned)(-1)) { 1540 /* Very unlikely, but possible on 16 bit machine if 1541 * strstart == 0 && lookahead == 1 (input done a byte at time) 1542 */ 1543 more--; 1544 } 1545 } 1546 1547 /* If the window is almost full and there is insufficient lookahead, 1548 * move the upper half to the lower one to make room in the upper half. 1549 */ 1550 if (s->strstart >= wsize+MAX_DIST(s)) { 1551 1552 zmemcpy(s->window, s->window+wsize, (unsigned)wsize - more); 1553 s->match_start -= wsize; 1554 s->strstart -= wsize; /* we now have strstart >= MAX_DIST */ 1555 s->block_start -= (long) wsize; 1556 if (s->insert > s->strstart) 1557 s->insert = s->strstart; 1558 slide_hash(s); 1559 more += wsize; 1560 } 1561 if (s->strm->avail_in == 0) break; 1562 1563 /* If there was no sliding: 1564 * strstart <= WSIZE+MAX_DIST-1 && lookahead <= MIN_LOOKAHEAD - 1 && 1565 * more == window_size - lookahead - strstart 1566 * => more >= window_size - (MIN_LOOKAHEAD-1 + WSIZE + MAX_DIST-1) 1567 * => more >= window_size - 2*WSIZE + 2 1568 * In the BIG_MEM or MMAP case (not yet supported), 1569 * window_size == input_size + MIN_LOOKAHEAD && 1570 * strstart + s->lookahead <= input_size => more >= MIN_LOOKAHEAD. 1571 * Otherwise, window_size == 2*WSIZE so more >= 2. 1572 * If there was sliding, more >= WSIZE. So in all cases, more >= 2. 1573 */ 1574 Assert(more >= 2, "more < 2"); 1575 1576 n = read_buf(s->strm, s->window + s->strstart + s->lookahead, more); 1577 s->lookahead += n; 1578 1579 /* Initialize the hash value now that we have some input: */ 1580 if (s->lookahead + s->insert >= MIN_MATCH) { 1581 uInt str = s->strstart - s->insert; 1582 s->ins_h = s->window[str]; 1583 UPDATE_HASH(s, s->ins_h, s->window[str + 1]); 1584 #if MIN_MATCH != 3 1585 Call UPDATE_HASH() MIN_MATCH-3 more times 1586 #endif 1587 while (s->insert) { 1588 UPDATE_HASH(s, s->ins_h, s->window[str + MIN_MATCH-1]); 1589 #ifndef FASTEST 1590 s->prev[str & s->w_mask] = s->head[s->ins_h]; 1591 #endif 1592 s->head[s->ins_h] = (Pos)str; 1593 str++; 1594 s->insert--; 1595 if (s->lookahead + s->insert < MIN_MATCH) 1596 break; 1597 } 1598 } 1599 /* If the whole input has less than MIN_MATCH bytes, ins_h is garbage, 1600 * but this is not important since only literal bytes will be emitted. 1601 */ 1602 1603 } while (s->lookahead < MIN_LOOKAHEAD && s->strm->avail_in != 0); 1604 1605 /* If the WIN_INIT bytes after the end of the current data have never been 1606 * written, then zero those bytes in order to avoid memory check reports of 1607 * the use of uninitialized (or uninitialised as Julian writes) bytes by 1608 * the longest match routines. Update the high water mark for the next 1609 * time through here. WIN_INIT is set to MAX_MATCH since the longest match 1610 * routines allow scanning to strstart + MAX_MATCH, ignoring lookahead. 1611 */ 1612 if (s->high_water < s->window_size) { 1613 ulg curr = s->strstart + (ulg)(s->lookahead); 1614 ulg init; 1615 1616 if (s->high_water < curr) { 1617 /* Previous high water mark below current data -- zero WIN_INIT 1618 * bytes or up to end of window, whichever is less. 1619 */ 1620 init = s->window_size - curr; 1621 if (init > WIN_INIT) 1622 init = WIN_INIT; 1623 zmemzero(s->window + curr, (unsigned)init); 1624 s->high_water = curr + init; 1625 } 1626 else if (s->high_water < (ulg)curr + WIN_INIT) { 1627 /* High water mark at or above current data, but below current data 1628 * plus WIN_INIT -- zero out to current data plus WIN_INIT, or up 1629 * to end of window, whichever is less. 1630 */ 1631 init = (ulg)curr + WIN_INIT - s->high_water; 1632 if (init > s->window_size - s->high_water) 1633 init = s->window_size - s->high_water; 1634 zmemzero(s->window + s->high_water, (unsigned)init); 1635 s->high_water += init; 1636 } 1637 } 1638 1639 Assert((ulg)s->strstart <= s->window_size - MIN_LOOKAHEAD, 1640 "not enough room for search"); 1641 } 1642 1643 /* =========================================================================== 1644 * Flush the current block, with given end-of-file flag. 1645 * IN assertion: strstart is set to the end of the current match. 1646 */ 1647 #define FLUSH_BLOCK_ONLY(s, last) { \ 1648 _tr_flush_block(s, (s->block_start >= 0L ? \ 1649 (charf *)&s->window[(unsigned)s->block_start] : \ 1650 (charf *)Z_NULL), \ 1651 (ulg)((long)s->strstart - s->block_start), \ 1652 (last)); \ 1653 s->block_start = s->strstart; \ 1654 flush_pending(s->strm); \ 1655 Tracev((stderr,"[FLUSH]")); \ 1656 } 1657 1658 /* Same but force premature exit if necessary. */ 1659 #define FLUSH_BLOCK(s, last) { \ 1660 FLUSH_BLOCK_ONLY(s, last); \ 1661 if (s->strm->avail_out == 0) return (last) ? finish_started : need_more; \ 1662 } 1663 1664 /* Maximum stored block length in deflate format (not including header). */ 1665 #define MAX_STORED 65535 1666 1667 #if !defined(MIN) 1668 /* Minimum of a and b. */ 1669 #define MIN(a, b) ((a) > (b) ? (b) : (a)) 1670 #endif 1671 1672 /* =========================================================================== 1673 * Copy without compression as much as possible from the input stream, return 1674 * the current block state. 1675 * 1676 * In case deflateParams() is used to later switch to a non-zero compression 1677 * level, s->matches (otherwise unused when storing) keeps track of the number 1678 * of hash table slides to perform. If s->matches is 1, then one hash table 1679 * slide will be done when switching. If s->matches is 2, the maximum value 1680 * allowed here, then the hash table will be cleared, since two or more slides 1681 * is the same as a clear. 1682 * 1683 * deflate_stored() is written to minimize the number of times an input byte is 1684 * copied. It is most efficient with large input and output buffers, which 1685 * maximizes the opportunites to have a single copy from next_in to next_out. 1686 */ 1687 local block_state deflate_stored(s, flush) 1688 deflate_state *s; 1689 int flush; 1690 { 1691 /* Smallest worthy block size when not flushing or finishing. By default 1692 * this is 32K. This can be as small as 507 bytes for memLevel == 1. For 1693 * large input and output buffers, the stored block size will be larger. 1694 */ 1695 unsigned min_block = MIN(s->pending_buf_size - 5, s->w_size); 1696 1697 /* Copy as many min_block or larger stored blocks directly to next_out as 1698 * possible. If flushing, copy the remaining available input to next_out as 1699 * stored blocks, if there is enough space. 1700 */ 1701 unsigned len, left, have, last = 0; 1702 unsigned used = s->strm->avail_in; 1703 do { 1704 /* Set len to the maximum size block that we can copy directly with the 1705 * available input data and output space. Set left to how much of that 1706 * would be copied from what's left in the window. 1707 */ 1708 len = MAX_STORED; /* maximum deflate stored block length */ 1709 have = (s->bi_valid + 42) >> 3; /* number of header bytes */ 1710 if (s->strm->avail_out < have) /* need room for header */ 1711 break; 1712 /* maximum stored block length that will fit in avail_out: */ 1713 have = s->strm->avail_out - have; 1714 left = s->strstart - s->block_start; /* bytes left in window */ 1715 if (len > (ulg)left + s->strm->avail_in) 1716 len = left + s->strm->avail_in; /* limit len to the input */ 1717 if (len > have) 1718 len = have; /* limit len to the output */ 1719 1720 /* If the stored block would be less than min_block in length, or if 1721 * unable to copy all of the available input when flushing, then try 1722 * copying to the window and the pending buffer instead. Also don't 1723 * write an empty block when flushing -- deflate() does that. 1724 */ 1725 if (len < min_block && ((len == 0 && flush != Z_FINISH) || 1726 flush == Z_NO_FLUSH || 1727 len != left + s->strm->avail_in)) 1728 break; 1729 1730 /* Make a dummy stored block in pending to get the header bytes, 1731 * including any pending bits. This also updates the debugging counts. 1732 */ 1733 last = flush == Z_FINISH && len == left + s->strm->avail_in ? 1 : 0; 1734 _tr_stored_block(s, (char *)0, 0L, last); 1735 1736 /* Replace the lengths in the dummy stored block with len. */ 1737 s->pending_buf[s->pending - 4] = len; 1738 s->pending_buf[s->pending - 3] = len >> 8; 1739 s->pending_buf[s->pending - 2] = ~len; 1740 s->pending_buf[s->pending - 1] = ~len >> 8; 1741 1742 /* Write the stored block header bytes. */ 1743 flush_pending(s->strm); 1744 1745 #ifdef ZLIB_DEBUG 1746 /* Update debugging counts for the data about to be copied. */ 1747 s->compressed_len += len << 3; 1748 s->bits_sent += len << 3; 1749 #endif 1750 1751 /* Copy uncompressed bytes from the window to next_out. */ 1752 if (left) { 1753 if (left > len) 1754 left = len; 1755 zmemcpy(s->strm->next_out, s->window + s->block_start, left); 1756 s->strm->next_out += left; 1757 s->strm->avail_out -= left; 1758 s->strm->total_out += left; 1759 s->block_start += left; 1760 len -= left; 1761 } 1762 1763 /* Copy uncompressed bytes directly from next_in to next_out, updating 1764 * the check value. 1765 */ 1766 if (len) { 1767 read_buf(s->strm, s->strm->next_out, len); 1768 s->strm->next_out += len; 1769 s->strm->avail_out -= len; 1770 s->strm->total_out += len; 1771 } 1772 } while (last == 0); 1773 1774 /* Update the sliding window with the last s->w_size bytes of the copied 1775 * data, or append all of the copied data to the existing window if less 1776 * than s->w_size bytes were copied. Also update the number of bytes to 1777 * insert in the hash tables, in the event that deflateParams() switches to 1778 * a non-zero compression level. 1779 */ 1780 used -= s->strm->avail_in; /* number of input bytes directly copied */ 1781 if (used) { 1782 /* If any input was used, then no unused input remains in the window, 1783 * therefore s->block_start == s->strstart. 1784 */ 1785 if (used >= s->w_size) { /* supplant the previous history */ 1786 s->matches = 2; /* clear hash */ 1787 zmemcpy(s->window, s->strm->next_in - s->w_size, s->w_size); 1788 s->strstart = s->w_size; 1789 s->insert = s->strstart; 1790 } 1791 else { 1792 if (s->window_size - s->strstart <= used) { 1793 /* Slide the window down. */ 1794 s->strstart -= s->w_size; 1795 zmemcpy(s->window, s->window + s->w_size, s->strstart); 1796 if (s->matches < 2) 1797 s->matches++; /* add a pending slide_hash() */ 1798 if (s->insert > s->strstart) 1799 s->insert = s->strstart; 1800 } 1801 zmemcpy(s->window + s->strstart, s->strm->next_in - used, used); 1802 s->strstart += used; 1803 s->insert += MIN(used, s->w_size - s->insert); 1804 } 1805 s->block_start = s->strstart; 1806 } 1807 if (s->high_water < s->strstart) 1808 s->high_water = s->strstart; 1809 1810 /* If the last block was written to next_out, then done. */ 1811 if (last) 1812 return finish_done; 1813 1814 /* If flushing and all input has been consumed, then done. */ 1815 if (flush != Z_NO_FLUSH && flush != Z_FINISH && 1816 s->strm->avail_in == 0 && (long)s->strstart == s->block_start) 1817 return block_done; 1818 1819 /* Fill the window with any remaining input. */ 1820 have = s->window_size - s->strstart; 1821 if (s->strm->avail_in > have && s->block_start >= (long)s->w_size) { 1822 /* Slide the window down. */ 1823 s->block_start -= s->w_size; 1824 s->strstart -= s->w_size; 1825 zmemcpy(s->window, s->window + s->w_size, s->strstart); 1826 if (s->matches < 2) 1827 s->matches++; /* add a pending slide_hash() */ 1828 have += s->w_size; /* more space now */ 1829 if (s->insert > s->strstart) 1830 s->insert = s->strstart; 1831 } 1832 if (have > s->strm->avail_in) 1833 have = s->strm->avail_in; 1834 if (have) { 1835 read_buf(s->strm, s->window + s->strstart, have); 1836 s->strstart += have; 1837 s->insert += MIN(have, s->w_size - s->insert); 1838 } 1839 if (s->high_water < s->strstart) 1840 s->high_water = s->strstart; 1841 1842 /* There was not enough avail_out to write a complete worthy or flushed 1843 * stored block to next_out. Write a stored block to pending instead, if we 1844 * have enough input for a worthy block, or if flushing and there is enough 1845 * room for the remaining input as a stored block in the pending buffer. 1846 */ 1847 have = (s->bi_valid + 42) >> 3; /* number of header bytes */ 1848 /* maximum stored block length that will fit in pending: */ 1849 have = MIN(s->pending_buf_size - have, MAX_STORED); 1850 min_block = MIN(have, s->w_size); 1851 left = s->strstart - s->block_start; 1852 if (left >= min_block || 1853 ((left || flush == Z_FINISH) && flush != Z_NO_FLUSH && 1854 s->strm->avail_in == 0 && left <= have)) { 1855 len = MIN(left, have); 1856 last = flush == Z_FINISH && s->strm->avail_in == 0 && 1857 len == left ? 1 : 0; 1858 _tr_stored_block(s, (charf *)s->window + s->block_start, len, last); 1859 s->block_start += len; 1860 flush_pending(s->strm); 1861 } 1862 1863 /* We've done all we can with the available input and output. */ 1864 return last ? finish_started : need_more; 1865 } 1866 1867 /* =========================================================================== 1868 * Compress as much as possible from the input stream, return the current 1869 * block state. 1870 * This function does not perform lazy evaluation of matches and inserts 1871 * new strings in the dictionary only for unmatched strings or for short 1872 * matches. It is used only for the fast compression options. 1873 */ 1874 local block_state deflate_fast(s, flush) 1875 deflate_state *s; 1876 int flush; 1877 { 1878 IPos hash_head; /* head of the hash chain */ 1879 int bflush; /* set if current block must be flushed */ 1880 1881 for (;;) { 1882 /* Make sure that we always have enough lookahead, except 1883 * at the end of the input file. We need MAX_MATCH bytes 1884 * for the next match, plus MIN_MATCH bytes to insert the 1885 * string following the next match. 1886 */ 1887 if (s->lookahead < MIN_LOOKAHEAD) { 1888 fill_window(s); 1889 if (s->lookahead < MIN_LOOKAHEAD && flush == Z_NO_FLUSH) { 1890 return need_more; 1891 } 1892 if (s->lookahead == 0) break; /* flush the current block */ 1893 } 1894 1895 /* Insert the string window[strstart .. strstart+2] in the 1896 * dictionary, and set hash_head to the head of the hash chain: 1897 */ 1898 hash_head = NIL; 1899 if (s->lookahead >= MIN_MATCH) { 1900 INSERT_STRING(s, s->strstart, hash_head); 1901 } 1902 1903 /* Find the longest match, discarding those <= prev_length. 1904 * At this point we have always match_length < MIN_MATCH 1905 */ 1906 if (hash_head != NIL && s->strstart - hash_head <= MAX_DIST(s)) { 1907 /* To simplify the code, we prevent matches with the string 1908 * of window index 0 (in particular we have to avoid a match 1909 * of the string with itself at the start of the input file). 1910 */ 1911 s->match_length = longest_match (s, hash_head); 1912 /* longest_match() sets match_start */ 1913 } 1914 if (s->match_length >= MIN_MATCH) { 1915 check_match(s, s->strstart, s->match_start, s->match_length); 1916 1917 _tr_tally_dist(s, s->strstart - s->match_start, 1918 s->match_length - MIN_MATCH, bflush); 1919 1920 s->lookahead -= s->match_length; 1921 1922 /* Insert new strings in the hash table only if the match length 1923 * is not too large. This saves time but degrades compression. 1924 */ 1925 #ifndef FASTEST 1926 if (s->match_length <= s->max_insert_length && 1927 s->lookahead >= MIN_MATCH) { 1928 s->match_length--; /* string at strstart already in table */ 1929 do { 1930 s->strstart++; 1931 INSERT_STRING(s, s->strstart, hash_head); 1932 /* strstart never exceeds WSIZE-MAX_MATCH, so there are 1933 * always MIN_MATCH bytes ahead. 1934 */ 1935 } while (--s->match_length != 0); 1936 s->strstart++; 1937 } else 1938 #endif 1939 { 1940 s->strstart += s->match_length; 1941 s->match_length = 0; 1942 s->ins_h = s->window[s->strstart]; 1943 UPDATE_HASH(s, s->ins_h, s->window[s->strstart+1]); 1944 #if MIN_MATCH != 3 1945 Call UPDATE_HASH() MIN_MATCH-3 more times 1946 #endif 1947 /* If lookahead < MIN_MATCH, ins_h is garbage, but it does not 1948 * matter since it will be recomputed at next deflate call. 1949 */ 1950 } 1951 } else { 1952 /* No match, output a literal byte */ 1953 Tracevv((stderr,"%c", s->window[s->strstart])); 1954 _tr_tally_lit (s, s->window[s->strstart], bflush); 1955 s->lookahead--; 1956 s->strstart++; 1957 } 1958 if (bflush) FLUSH_BLOCK(s, 0); 1959 } 1960 s->insert = s->strstart < MIN_MATCH-1 ? s->strstart : MIN_MATCH-1; 1961 if (flush == Z_FINISH) { 1962 FLUSH_BLOCK(s, 1); 1963 return finish_done; 1964 } 1965 if (s->sym_next) 1966 FLUSH_BLOCK(s, 0); 1967 return block_done; 1968 } 1969 1970 #ifndef FASTEST 1971 /* =========================================================================== 1972 * Same as above, but achieves better compression. We use a lazy 1973 * evaluation for matches: a match is finally adopted only if there is 1974 * no better match at the next window position. 1975 */ 1976 local block_state deflate_slow(s, flush) 1977 deflate_state *s; 1978 int flush; 1979 { 1980 IPos hash_head; /* head of hash chain */ 1981 int bflush; /* set if current block must be flushed */ 1982 1983 /* Process the input block. */ 1984 for (;;) { 1985 /* Make sure that we always have enough lookahead, except 1986 * at the end of the input file. We need MAX_MATCH bytes 1987 * for the next match, plus MIN_MATCH bytes to insert the 1988 * string following the next match. 1989 */ 1990 if (s->lookahead < MIN_LOOKAHEAD) { 1991 fill_window(s); 1992 if (s->lookahead < MIN_LOOKAHEAD && flush == Z_NO_FLUSH) { 1993 return need_more; 1994 } 1995 if (s->lookahead == 0) break; /* flush the current block */ 1996 } 1997 1998 /* Insert the string window[strstart .. strstart+2] in the 1999 * dictionary, and set hash_head to the head of the hash chain: 2000 */ 2001 hash_head = NIL; 2002 if (s->lookahead >= MIN_MATCH) { 2003 INSERT_STRING(s, s->strstart, hash_head); 2004 } 2005 2006 /* Find the longest match, discarding those <= prev_length. 2007 */ 2008 s->prev_length = s->match_length, s->prev_match = s->match_start; 2009 s->match_length = MIN_MATCH-1; 2010 2011 if (hash_head != NIL && s->prev_length < s->max_lazy_match && 2012 s->strstart - hash_head <= MAX_DIST(s)) { 2013 /* To simplify the code, we prevent matches with the string 2014 * of window index 0 (in particular we have to avoid a match 2015 * of the string with itself at the start of the input file). 2016 */ 2017 s->match_length = longest_match (s, hash_head); 2018 /* longest_match() sets match_start */ 2019 2020 if (s->match_length <= 5 && (s->strategy == Z_FILTERED 2021 #if TOO_FAR <= 32767 2022 || (s->match_length == MIN_MATCH && 2023 s->strstart - s->match_start > TOO_FAR) 2024 #endif 2025 )) { 2026 2027 /* If prev_match is also MIN_MATCH, match_start is garbage 2028 * but we will ignore the current match anyway. 2029 */ 2030 s->match_length = MIN_MATCH-1; 2031 } 2032 } 2033 /* If there was a match at the previous step and the current 2034 * match is not better, output the previous match: 2035 */ 2036 if (s->prev_length >= MIN_MATCH && s->match_length <= s->prev_length) { 2037 uInt max_insert = s->strstart + s->lookahead - MIN_MATCH; 2038 /* Do not insert strings in hash table beyond this. */ 2039 2040 check_match(s, s->strstart-1, s->prev_match, s->prev_length); 2041 2042 _tr_tally_dist(s, s->strstart -1 - s->prev_match, 2043 s->prev_length - MIN_MATCH, bflush); 2044 2045 /* Insert in hash table all strings up to the end of the match. 2046 * strstart-1 and strstart are already inserted. If there is not 2047 * enough lookahead, the last two strings are not inserted in 2048 * the hash table. 2049 */ 2050 s->lookahead -= s->prev_length-1; 2051 s->prev_length -= 2; 2052 do { 2053 if (++s->strstart <= max_insert) { 2054 INSERT_STRING(s, s->strstart, hash_head); 2055 } 2056 } while (--s->prev_length != 0); 2057 s->match_available = 0; 2058 s->match_length = MIN_MATCH-1; 2059 s->strstart++; 2060 2061 if (bflush) FLUSH_BLOCK(s, 0); 2062 2063 } else if (s->match_available) { 2064 /* If there was no match at the previous position, output a 2065 * single literal. If there was a match but the current match 2066 * is longer, truncate the previous match to a single literal. 2067 */ 2068 Tracevv((stderr,"%c", s->window[s->strstart-1])); 2069 _tr_tally_lit(s, s->window[s->strstart-1], bflush); 2070 if (bflush) { 2071 FLUSH_BLOCK_ONLY(s, 0); 2072 } 2073 s->strstart++; 2074 s->lookahead--; 2075 if (s->strm->avail_out == 0) return need_more; 2076 } else { 2077 /* There is no previous match to compare with, wait for 2078 * the next step to decide. 2079 */ 2080 s->match_available = 1; 2081 s->strstart++; 2082 s->lookahead--; 2083 } 2084 } 2085 Assert (flush != Z_NO_FLUSH, "no flush?"); 2086 if (s->match_available) { 2087 Tracevv((stderr,"%c", s->window[s->strstart-1])); 2088 _tr_tally_lit(s, s->window[s->strstart-1], bflush); 2089 s->match_available = 0; 2090 } 2091 s->insert = s->strstart < MIN_MATCH-1 ? s->strstart : MIN_MATCH-1; 2092 if (flush == Z_FINISH) { 2093 FLUSH_BLOCK(s, 1); 2094 return finish_done; 2095 } 2096 if (s->sym_next) 2097 FLUSH_BLOCK(s, 0); 2098 return block_done; 2099 } 2100 #endif /* FASTEST */ 2101 2102 /* =========================================================================== 2103 * For Z_RLE, simply look for runs of bytes, generate matches only of distance 2104 * one. Do not maintain a hash table. (It will be regenerated if this run of 2105 * deflate switches away from Z_RLE.) 2106 */ 2107 local block_state deflate_rle(s, flush) 2108 deflate_state *s; 2109 int flush; 2110 { 2111 int bflush; /* set if current block must be flushed */ 2112 uInt prev; /* byte at distance one to match */ 2113 Bytef *scan, *strend; /* scan goes up to strend for length of run */ 2114 2115 for (;;) { 2116 /* Make sure that we always have enough lookahead, except 2117 * at the end of the input file. We need MAX_MATCH bytes 2118 * for the longest run, plus one for the unrolled loop. 2119 */ 2120 if (s->lookahead <= MAX_MATCH) { 2121 fill_window(s); 2122 if (s->lookahead <= MAX_MATCH && flush == Z_NO_FLUSH) { 2123 return need_more; 2124 } 2125 if (s->lookahead == 0) break; /* flush the current block */ 2126 } 2127 2128 /* See how many times the previous byte repeats */ 2129 s->match_length = 0; 2130 if (s->lookahead >= MIN_MATCH && s->strstart > 0) { 2131 scan = s->window + s->strstart - 1; 2132 prev = *scan; 2133 if (prev == *++scan && prev == *++scan && prev == *++scan) { 2134 strend = s->window + s->strstart + MAX_MATCH; 2135 do { 2136 } while (prev == *++scan && prev == *++scan && 2137 prev == *++scan && prev == *++scan && 2138 prev == *++scan && prev == *++scan && 2139 prev == *++scan && prev == *++scan && 2140 scan < strend); 2141 s->match_length = MAX_MATCH - (uInt)(strend - scan); 2142 if (s->match_length > s->lookahead) 2143 s->match_length = s->lookahead; 2144 } 2145 Assert(scan <= s->window+(uInt)(s->window_size-1), "wild scan"); 2146 } 2147 2148 /* Emit match if have run of MIN_MATCH or longer, else emit literal */ 2149 if (s->match_length >= MIN_MATCH) { 2150 check_match(s, s->strstart, s->strstart - 1, s->match_length); 2151 2152 _tr_tally_dist(s, 1, s->match_length - MIN_MATCH, bflush); 2153 2154 s->lookahead -= s->match_length; 2155 s->strstart += s->match_length; 2156 s->match_length = 0; 2157 } else { 2158 /* No match, output a literal byte */ 2159 Tracevv((stderr,"%c", s->window[s->strstart])); 2160 _tr_tally_lit (s, s->window[s->strstart], bflush); 2161 s->lookahead--; 2162 s->strstart++; 2163 } 2164 if (bflush) FLUSH_BLOCK(s, 0); 2165 } 2166 s->insert = 0; 2167 if (flush == Z_FINISH) { 2168 FLUSH_BLOCK(s, 1); 2169 return finish_done; 2170 } 2171 if (s->sym_next) 2172 FLUSH_BLOCK(s, 0); 2173 return block_done; 2174 } 2175 2176 /* =========================================================================== 2177 * For Z_HUFFMAN_ONLY, do not look for matches. Do not maintain a hash table. 2178 * (It will be regenerated if this run of deflate switches away from Huffman.) 2179 */ 2180 local block_state deflate_huff(s, flush) 2181 deflate_state *s; 2182 int flush; 2183 { 2184 int bflush; /* set if current block must be flushed */ 2185 2186 for (;;) { 2187 /* Make sure that we have a literal to write. */ 2188 if (s->lookahead == 0) { 2189 fill_window(s); 2190 if (s->lookahead == 0) { 2191 if (flush == Z_NO_FLUSH) 2192 return need_more; 2193 break; /* flush the current block */ 2194 } 2195 } 2196 2197 /* Output a literal byte */ 2198 s->match_length = 0; 2199 Tracevv((stderr,"%c", s->window[s->strstart])); 2200 _tr_tally_lit (s, s->window[s->strstart], bflush); 2201 s->lookahead--; 2202 s->strstart++; 2203 if (bflush) FLUSH_BLOCK(s, 0); 2204 } 2205 s->insert = 0; 2206 if (flush == Z_FINISH) { 2207 FLUSH_BLOCK(s, 1); 2208 return finish_done; 2209 } 2210 if (s->sym_next) 2211 FLUSH_BLOCK(s, 0); 2212 return block_done; 2213 } 2214