OLD | NEW |
| (Empty) |
1 /* $Id: tif_lzw.c,v 1.45 2011-04-02 20:54:09 bfriesen Exp $ */ | |
2 | |
3 /* | |
4 * Copyright (c) 1988-1997 Sam Leffler | |
5 * Copyright (c) 1991-1997 Silicon Graphics, Inc. | |
6 * | |
7 * Permission to use, copy, modify, distribute, and sell this software and | |
8 * its documentation for any purpose is hereby granted without fee, provided | |
9 * that (i) the above copyright notices and this permission notice appear in | |
10 * all copies of the software and related documentation, and (ii) the names of | |
11 * Sam Leffler and Silicon Graphics may not be used in any advertising or | |
12 * publicity relating to the software without the specific, prior written | |
13 * permission of Sam Leffler and Silicon Graphics. | |
14 * | |
15 * THE SOFTWARE IS PROVIDED "AS-IS" AND WITHOUT WARRANTY OF ANY KIND, | |
16 * EXPRESS, IMPLIED OR OTHERWISE, INCLUDING WITHOUT LIMITATION, ANY | |
17 * WARRANTY OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. | |
18 * | |
19 * IN NO EVENT SHALL SAM LEFFLER OR SILICON GRAPHICS BE LIABLE FOR | |
20 * ANY SPECIAL, INCIDENTAL, INDIRECT OR CONSEQUENTIAL DAMAGES OF ANY KIND, | |
21 * OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, | |
22 * WHETHER OR NOT ADVISED OF THE POSSIBILITY OF DAMAGE, AND ON ANY THEORY OF | |
23 * LIABILITY, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE | |
24 * OF THIS SOFTWARE. | |
25 */ | |
26 #include "tiffiop.h" | |
27 #ifdef LZW_SUPPORT | |
28 /* | |
29 * TIFF Library. | |
30 * Rev 5.0 Lempel-Ziv & Welch Compression Support | |
31 * | |
32 * This code is derived from the compress program whose code is | |
33 * derived from software contributed to Berkeley by James A. Woods, | |
34 * derived from original work by Spencer Thomas and Joseph Orost. | |
35 * | |
36 * The original Berkeley copyright notice appears below in its entirety. | |
37 */ | |
38 #include "tif_predict.h" | |
39 | |
40 #include <stdio.h> | |
41 | |
42 /* | |
43 * NB: The 5.0 spec describes a different algorithm than Aldus | |
44 * implements. Specifically, Aldus does code length transitions | |
45 * one code earlier than should be done (for real LZW). | |
46 * Earlier versions of this library implemented the correct | |
47 * LZW algorithm, but emitted codes in a bit order opposite | |
48 * to the TIFF spec. Thus, to maintain compatibility w/ Aldus | |
49 * we interpret MSB-LSB ordered codes to be images written w/ | |
50 * old versions of this library, but otherwise adhere to the | |
51 * Aldus "off by one" algorithm. | |
52 * | |
53 * Future revisions to the TIFF spec are expected to "clarify this issue". | |
54 */ | |
55 #define LZW_COMPAT /* include backwards compatibility code */ | |
56 /* | |
57 * Each strip of data is supposed to be terminated by a CODE_EOI. | |
58 * If the following #define is included, the decoder will also | |
59 * check for end-of-strip w/o seeing this code. This makes the | |
60 * library more robust, but also slower. | |
61 */ | |
62 #define LZW_CHECKEOS /* include checks for strips w/o EOI code */ | |
63 | |
64 #define MAXCODE(n) ((1L<<(n))-1) | |
65 /* | |
66 * The TIFF spec specifies that encoded bit | |
67 * strings range from 9 to 12 bits. | |
68 */ | |
69 #define BITS_MIN 9 /* start with 9 bits */ | |
70 #define BITS_MAX 12 /* max of 12 bit strings */ | |
71 /* predefined codes */ | |
72 #define CODE_CLEAR 256 /* code to clear string table */ | |
73 #define CODE_EOI 257 /* end-of-information code */ | |
74 #define CODE_FIRST 258 /* first free code entry */ | |
75 #define CODE_MAX MAXCODE(BITS_MAX) | |
76 #define HSIZE 9001L /* 91% occupancy */ | |
77 #define HSHIFT (13-8) | |
78 #ifdef LZW_COMPAT | |
79 /* NB: +1024 is for compatibility with old files */ | |
80 #define CSIZE (MAXCODE(BITS_MAX)+1024L) | |
81 #else | |
82 #define CSIZE (MAXCODE(BITS_MAX)+1L) | |
83 #endif | |
84 | |
85 /* | |
86 * State block for each open TIFF file using LZW | |
87 * compression/decompression. Note that the predictor | |
88 * state block must be first in this data structure. | |
89 */ | |
90 typedef struct { | |
91 TIFFPredictorState predict; /* predictor super class */ | |
92 | |
93 unsigned short nbits; /* # of bits/code */ | |
94 unsigned short maxcode; /* maximum code for lzw_nbits */ | |
95 unsigned short free_ent; /* next free entry in hash table */ | |
96 long nextdata; /* next bits of i/o */ | |
97 long nextbits; /* # of valid bits in lzw_nextdata */ | |
98 | |
99 int rw_mode; /* preserve rw_mode from init */ | |
100 } LZWBaseState; | |
101 | |
102 #define lzw_nbits base.nbits | |
103 #define lzw_maxcode base.maxcode | |
104 #define lzw_free_ent base.free_ent | |
105 #define lzw_nextdata base.nextdata | |
106 #define lzw_nextbits base.nextbits | |
107 | |
108 /* | |
109 * Encoding-specific state. | |
110 */ | |
111 typedef uint16 hcode_t; /* codes fit in 16 bits */ | |
112 typedef struct { | |
113 long hash; | |
114 hcode_t code; | |
115 } hash_t; | |
116 | |
117 /* | |
118 * Decoding-specific state. | |
119 */ | |
120 typedef struct code_ent { | |
121 struct code_ent *next; | |
122 unsigned short length; /* string len, including this token */ | |
123 unsigned char value; /* data value */ | |
124 unsigned char firstchar; /* first token of string */ | |
125 } code_t; | |
126 | |
127 typedef int (*decodeFunc)(TIFF*, uint8*, tmsize_t, uint16); | |
128 | |
129 typedef struct { | |
130 LZWBaseState base; | |
131 | |
132 /* Decoding specific data */ | |
133 long dec_nbitsmask; /* lzw_nbits 1 bits, right adjusted */ | |
134 long dec_restart; /* restart count */ | |
135 #ifdef LZW_CHECKEOS | |
136 uint64 dec_bitsleft; /* available bits in raw data */ | |
137 #endif | |
138 decodeFunc dec_decode; /* regular or backwards compatible */ | |
139 code_t* dec_codep; /* current recognized code */ | |
140 code_t* dec_oldcodep; /* previously recognized code */ | |
141 code_t* dec_free_entp; /* next free entry */ | |
142 code_t* dec_maxcodep; /* max available entry */ | |
143 code_t* dec_codetab; /* kept separate for small machines */ | |
144 | |
145 /* Encoding specific data */ | |
146 int enc_oldcode; /* last code encountered */ | |
147 long enc_checkpoint; /* point at which to clear table */ | |
148 #define CHECK_GAP 10000 /* enc_ratio check interval */ | |
149 long enc_ratio; /* current compression ratio */ | |
150 long enc_incount; /* (input) data bytes encoded */ | |
151 long enc_outcount; /* encoded (output) bytes */ | |
152 uint8* enc_rawlimit; /* bound on tif_rawdata buffer */ | |
153 hash_t* enc_hashtab; /* kept separate for small machines */ | |
154 } LZWCodecState; | |
155 | |
156 #define LZWState(tif) ((LZWBaseState*) (tif)->tif_data) | |
157 #define DecoderState(tif) ((LZWCodecState*) LZWState(tif)) | |
158 #define EncoderState(tif) ((LZWCodecState*) LZWState(tif)) | |
159 | |
160 static int LZWDecode(TIFF* tif, uint8* op0, tmsize_t occ0, uint16 s); | |
161 #ifdef LZW_COMPAT | |
162 static int LZWDecodeCompat(TIFF* tif, uint8* op0, tmsize_t occ0, uint16 s); | |
163 #endif | |
164 static void cl_hash(LZWCodecState*); | |
165 | |
166 /* | |
167 * LZW Decoder. | |
168 */ | |
169 | |
170 #ifdef LZW_CHECKEOS | |
171 /* | |
172 * This check shouldn't be necessary because each | |
173 * strip is suppose to be terminated with CODE_EOI. | |
174 */ | |
175 #define NextCode(_tif, _sp, _bp, _code, _get) { \ | |
176 if ((_sp)->dec_bitsleft < (uint64)nbits) { \ | |
177 TIFFWarningExt(_tif->tif_clientdata, module, \ | |
178 "LZWDecode: Strip %d not terminated with EOI code", \ | |
179 _tif->tif_curstrip); \ | |
180 _code = CODE_EOI; \ | |
181 } else { \ | |
182 _get(_sp,_bp,_code); \ | |
183 (_sp)->dec_bitsleft -= nbits; \ | |
184 } \ | |
185 } | |
186 #else | |
187 #define NextCode(tif, sp, bp, code, get) get(sp, bp, code) | |
188 #endif | |
189 | |
190 static int | |
191 LZWFixupTags(TIFF* tif) | |
192 { | |
193 (void) tif; | |
194 return (1); | |
195 } | |
196 | |
197 static int | |
198 LZWSetupDecode(TIFF* tif) | |
199 { | |
200 static const char module[] = "LZWSetupDecode"; | |
201 LZWCodecState* sp = DecoderState(tif); | |
202 int code; | |
203 | |
204 if( sp == NULL ) | |
205 { | |
206 /* | |
207 * Allocate state block so tag methods have storage to record | |
208 * values. | |
209 */ | |
210 tif->tif_data = (uint8*) _TIFFmalloc(sizeof(LZWCodecState)); | |
211 if (tif->tif_data == NULL) | |
212 { | |
213 TIFFErrorExt(tif->tif_clientdata, module, "No space for
LZW state block"); | |
214 return (0); | |
215 } | |
216 | |
217 DecoderState(tif)->dec_codetab = NULL; | |
218 DecoderState(tif)->dec_decode = NULL; | |
219 | |
220 /* | |
221 * Setup predictor setup. | |
222 */ | |
223 (void) TIFFPredictorInit(tif); | |
224 | |
225 sp = DecoderState(tif); | |
226 } | |
227 | |
228 assert(sp != NULL); | |
229 | |
230 if (sp->dec_codetab == NULL) { | |
231 sp->dec_codetab = (code_t*)_TIFFmalloc(CSIZE*sizeof (code_t)); | |
232 if (sp->dec_codetab == NULL) { | |
233 TIFFErrorExt(tif->tif_clientdata, module, | |
234 "No space for LZW code table"); | |
235 return (0); | |
236 } | |
237 /* | |
238 * Pre-load the table. | |
239 */ | |
240 code = 255; | |
241 do { | |
242 sp->dec_codetab[code].value = code; | |
243 sp->dec_codetab[code].firstchar = code; | |
244 sp->dec_codetab[code].length = 1; | |
245 sp->dec_codetab[code].next = NULL; | |
246 } while (code--); | |
247 /* | |
248 * Zero-out the unused entries | |
249 */ | |
250 _TIFFmemset(&sp->dec_codetab[CODE_CLEAR], 0, | |
251 (CODE_FIRST - CODE_CLEAR) * sizeof (code_t)); | |
252 } | |
253 return (1); | |
254 } | |
255 | |
256 /* | |
257 * Setup state for decoding a strip. | |
258 */ | |
259 static int | |
260 LZWPreDecode(TIFF* tif, uint16 s) | |
261 { | |
262 static const char module[] = "LZWPreDecode"; | |
263 LZWCodecState *sp = DecoderState(tif); | |
264 | |
265 (void) s; | |
266 assert(sp != NULL); | |
267 if( sp->dec_codetab == NULL ) | |
268 { | |
269 tif->tif_setupdecode( tif ); | |
270 } | |
271 | |
272 /* | |
273 * Check for old bit-reversed codes. | |
274 */ | |
275 if (tif->tif_rawdata[0] == 0 && (tif->tif_rawdata[1] & 0x1)) { | |
276 #ifdef LZW_COMPAT | |
277 if (!sp->dec_decode) { | |
278 TIFFWarningExt(tif->tif_clientdata, module, | |
279 "Old-style LZW codes, convert file"); | |
280 /* | |
281 * Override default decoding methods with | |
282 * ones that deal with the old coding. | |
283 * Otherwise the predictor versions set | |
284 * above will call the compatibility routines | |
285 * through the dec_decode method. | |
286 */ | |
287 tif->tif_decoderow = LZWDecodeCompat; | |
288 tif->tif_decodestrip = LZWDecodeCompat; | |
289 tif->tif_decodetile = LZWDecodeCompat; | |
290 /* | |
291 * If doing horizontal differencing, must | |
292 * re-setup the predictor logic since we | |
293 * switched the basic decoder methods... | |
294 */ | |
295 (*tif->tif_setupdecode)(tif); | |
296 sp->dec_decode = LZWDecodeCompat; | |
297 } | |
298 sp->lzw_maxcode = MAXCODE(BITS_MIN); | |
299 #else /* !LZW_COMPAT */ | |
300 if (!sp->dec_decode) { | |
301 TIFFErrorExt(tif->tif_clientdata, module, | |
302 "Old-style LZW codes not supported"); | |
303 sp->dec_decode = LZWDecode; | |
304 } | |
305 return (0); | |
306 #endif/* !LZW_COMPAT */ | |
307 } else { | |
308 sp->lzw_maxcode = MAXCODE(BITS_MIN)-1; | |
309 sp->dec_decode = LZWDecode; | |
310 } | |
311 sp->lzw_nbits = BITS_MIN; | |
312 sp->lzw_nextbits = 0; | |
313 sp->lzw_nextdata = 0; | |
314 | |
315 sp->dec_restart = 0; | |
316 sp->dec_nbitsmask = MAXCODE(BITS_MIN); | |
317 #ifdef LZW_CHECKEOS | |
318 sp->dec_bitsleft = ((uint64)tif->tif_rawcc) << 3; | |
319 #endif | |
320 sp->dec_free_entp = sp->dec_codetab + CODE_FIRST; | |
321 /* | |
322 * Zero entries that are not yet filled in. We do | |
323 * this to guard against bogus input data that causes | |
324 * us to index into undefined entries. If you can | |
325 * come up with a way to safely bounds-check input codes | |
326 * while decoding then you can remove this operation. | |
327 */ | |
328 _TIFFmemset(sp->dec_free_entp, 0, (CSIZE-CODE_FIRST)*sizeof (code_t)); | |
329 sp->dec_oldcodep = &sp->dec_codetab[-1]; | |
330 sp->dec_maxcodep = &sp->dec_codetab[sp->dec_nbitsmask-1]; | |
331 return (1); | |
332 } | |
333 | |
334 /* | |
335 * Decode a "hunk of data". | |
336 */ | |
337 #define GetNextCode(sp, bp, code) { \ | |
338 nextdata = (nextdata<<8) | *(bp)++; \ | |
339 nextbits += 8; \ | |
340 if (nextbits < nbits) { \ | |
341 nextdata = (nextdata<<8) | *(bp)++; \ | |
342 nextbits += 8; \ | |
343 } \ | |
344 code = (hcode_t)((nextdata >> (nextbits-nbits)) & nbitsmask); \ | |
345 nextbits -= nbits; \ | |
346 } | |
347 | |
348 static void | |
349 codeLoop(TIFF* tif, const char* module) | |
350 { | |
351 TIFFErrorExt(tif->tif_clientdata, module, | |
352 "Bogus encoding, loop in the code table; scanline %d", | |
353 tif->tif_row); | |
354 } | |
355 | |
356 static int | |
357 LZWDecode(TIFF* tif, uint8* op0, tmsize_t occ0, uint16 s) | |
358 { | |
359 static const char module[] = "LZWDecode"; | |
360 LZWCodecState *sp = DecoderState(tif); | |
361 char *op = (char*) op0; | |
362 long occ = (long) occ0; | |
363 char *tp; | |
364 unsigned char *bp; | |
365 hcode_t code; | |
366 int len; | |
367 long nbits, nextbits, nextdata, nbitsmask; | |
368 code_t *codep, *free_entp, *maxcodep, *oldcodep; | |
369 | |
370 (void) s; | |
371 assert(sp != NULL); | |
372 assert(sp->dec_codetab != NULL); | |
373 | |
374 /* | |
375 Fail if value does not fit in long. | |
376 */ | |
377 if ((tmsize_t) occ != occ0) | |
378 return (0); | |
379 /* | |
380 * Restart interrupted output operation. | |
381 */ | |
382 if (sp->dec_restart) { | |
383 long residue; | |
384 | |
385 codep = sp->dec_codep; | |
386 residue = codep->length - sp->dec_restart; | |
387 if (residue > occ) { | |
388 /* | |
389 * Residue from previous decode is sufficient | |
390 * to satisfy decode request. Skip to the | |
391 * start of the decoded string, place decoded | |
392 * values in the output buffer, and return. | |
393 */ | |
394 sp->dec_restart += occ; | |
395 do { | |
396 codep = codep->next; | |
397 } while (--residue > occ && codep); | |
398 if (codep) { | |
399 tp = op + occ; | |
400 do { | |
401 *--tp = codep->value; | |
402 codep = codep->next; | |
403 } while (--occ && codep); | |
404 } | |
405 return (1); | |
406 } | |
407 /* | |
408 * Residue satisfies only part of the decode request. | |
409 */ | |
410 op += residue, occ -= residue; | |
411 tp = op; | |
412 do { | |
413 int t; | |
414 --tp; | |
415 t = codep->value; | |
416 codep = codep->next; | |
417 *tp = t; | |
418 } while (--residue && codep); | |
419 sp->dec_restart = 0; | |
420 } | |
421 | |
422 bp = (unsigned char *)tif->tif_rawcp; | |
423 nbits = sp->lzw_nbits; | |
424 nextdata = sp->lzw_nextdata; | |
425 nextbits = sp->lzw_nextbits; | |
426 nbitsmask = sp->dec_nbitsmask; | |
427 oldcodep = sp->dec_oldcodep; | |
428 free_entp = sp->dec_free_entp; | |
429 maxcodep = sp->dec_maxcodep; | |
430 | |
431 while (occ > 0) { | |
432 NextCode(tif, sp, bp, code, GetNextCode); | |
433 if (code == CODE_EOI) | |
434 break; | |
435 if (code == CODE_CLEAR) { | |
436 free_entp = sp->dec_codetab + CODE_FIRST; | |
437 _TIFFmemset(free_entp, 0, | |
438 (CSIZE - CODE_FIRST) * sizeof (code_t)); | |
439 nbits = BITS_MIN; | |
440 nbitsmask = MAXCODE(BITS_MIN); | |
441 maxcodep = sp->dec_codetab + nbitsmask-1; | |
442 NextCode(tif, sp, bp, code, GetNextCode); | |
443 if (code == CODE_EOI) | |
444 break; | |
445 if (code >= CODE_CLEAR) { | |
446 TIFFErrorExt(tif->tif_clientdata, tif->tif_name, | |
447 "LZWDecode: Corrupted LZW table at scanline %d", | |
448 tif->tif_row); | |
449 return (0); | |
450 } | |
451 *op++ = (char)code, occ--; | |
452 oldcodep = sp->dec_codetab + code; | |
453 continue; | |
454 } | |
455 codep = sp->dec_codetab + code; | |
456 | |
457 /* | |
458 * Add the new entry to the code table. | |
459 */ | |
460 if (free_entp < &sp->dec_codetab[0] || | |
461 free_entp >= &sp->dec_codetab[CSIZE]) { | |
462 TIFFErrorExt(tif->tif_clientdata, module, | |
463 "Corrupted LZW table at scanline %d", | |
464 tif->tif_row); | |
465 return (0); | |
466 } | |
467 | |
468 free_entp->next = oldcodep; | |
469 if (free_entp->next < &sp->dec_codetab[0] || | |
470 free_entp->next >= &sp->dec_codetab[CSIZE]) { | |
471 TIFFErrorExt(tif->tif_clientdata, module, | |
472 "Corrupted LZW table at scanline %d", | |
473 tif->tif_row); | |
474 return (0); | |
475 } | |
476 free_entp->firstchar = free_entp->next->firstchar; | |
477 free_entp->length = free_entp->next->length+1; | |
478 free_entp->value = (codep < free_entp) ? | |
479 codep->firstchar : free_entp->firstchar; | |
480 if (++free_entp > maxcodep) { | |
481 if (++nbits > BITS_MAX) /* should not happen */ | |
482 nbits = BITS_MAX; | |
483 nbitsmask = MAXCODE(nbits); | |
484 maxcodep = sp->dec_codetab + nbitsmask-1; | |
485 } | |
486 oldcodep = codep; | |
487 if (code >= 256) { | |
488 /* | |
489 * Code maps to a string, copy string | |
490 * value to output (written in reverse). | |
491 */ | |
492 if(codep->length == 0) { | |
493 TIFFErrorExt(tif->tif_clientdata, module, | |
494 "Wrong length of decoded string: " | |
495 "data probably corrupted at scanline %d", | |
496 tif->tif_row); | |
497 return (0); | |
498 } | |
499 if (codep->length > occ) { | |
500 /* | |
501 * String is too long for decode buffer, | |
502 * locate portion that will fit, copy to | |
503 * the decode buffer, and setup restart | |
504 * logic for the next decoding call. | |
505 */ | |
506 sp->dec_codep = codep; | |
507 do { | |
508 codep = codep->next; | |
509 } while (codep && codep->length > occ); | |
510 if (codep) { | |
511 sp->dec_restart = (long)occ; | |
512 tp = op + occ; | |
513 do { | |
514 *--tp = codep->value; | |
515 codep = codep->next; | |
516 } while (--occ && codep); | |
517 if (codep) | |
518 codeLoop(tif, module); | |
519 } | |
520 break; | |
521 } | |
522 len = codep->length; | |
523 tp = op + len; | |
524 do { | |
525 int t; | |
526 --tp; | |
527 t = codep->value; | |
528 codep = codep->next; | |
529 *tp = t; | |
530 } while (codep && tp > op); | |
531 if (codep) { | |
532 codeLoop(tif, module); | |
533 break; | |
534 } | |
535 assert(occ >= len); | |
536 op += len, occ -= len; | |
537 } else | |
538 *op++ = (char)code, occ--; | |
539 } | |
540 | |
541 tif->tif_rawcp = (uint8*) bp; | |
542 sp->lzw_nbits = (unsigned short) nbits; | |
543 sp->lzw_nextdata = nextdata; | |
544 sp->lzw_nextbits = nextbits; | |
545 sp->dec_nbitsmask = nbitsmask; | |
546 sp->dec_oldcodep = oldcodep; | |
547 sp->dec_free_entp = free_entp; | |
548 sp->dec_maxcodep = maxcodep; | |
549 | |
550 if (occ > 0) { | |
551 #if defined(__WIN32__) && (defined(_MSC_VER) || defined(__MINGW32__)) | |
552 TIFFErrorExt(tif->tif_clientdata, module, | |
553 "Not enough data at scanline %d (short %I64d bytes)", | |
554 tif->tif_row, (unsigned __int64) occ); | |
555 #else | |
556 TIFFErrorExt(tif->tif_clientdata, module, | |
557 "Not enough data at scanline %d (short %llu bytes)", | |
558 tif->tif_row, (unsigned long long) occ); | |
559 #endif | |
560 return (0); | |
561 } | |
562 return (1); | |
563 } | |
564 | |
565 #ifdef LZW_COMPAT | |
566 /* | |
567 * Decode a "hunk of data" for old images. | |
568 */ | |
569 #define GetNextCodeCompat(sp, bp, code) { \ | |
570 nextdata |= (unsigned long) *(bp)++ << nextbits; \ | |
571 nextbits += 8; \ | |
572 if (nextbits < nbits) { \ | |
573 nextdata |= (unsigned long) *(bp)++ << nextbits;\ | |
574 nextbits += 8; \ | |
575 } \ | |
576 code = (hcode_t)(nextdata & nbitsmask); \ | |
577 nextdata >>= nbits; \ | |
578 nextbits -= nbits; \ | |
579 } | |
580 | |
581 static int | |
582 LZWDecodeCompat(TIFF* tif, uint8* op0, tmsize_t occ0, uint16 s) | |
583 { | |
584 static const char module[] = "LZWDecodeCompat"; | |
585 LZWCodecState *sp = DecoderState(tif); | |
586 char *op = (char*) op0; | |
587 long occ = (long) occ0; | |
588 char *tp; | |
589 unsigned char *bp; | |
590 int code, nbits; | |
591 long nextbits, nextdata, nbitsmask; | |
592 code_t *codep, *free_entp, *maxcodep, *oldcodep; | |
593 | |
594 (void) s; | |
595 assert(sp != NULL); | |
596 | |
597 /* | |
598 Fail if value does not fit in long. | |
599 */ | |
600 if ((tmsize_t) occ != occ0) | |
601 return (0); | |
602 | |
603 /* | |
604 * Restart interrupted output operation. | |
605 */ | |
606 if (sp->dec_restart) { | |
607 long residue; | |
608 | |
609 codep = sp->dec_codep; | |
610 residue = codep->length - sp->dec_restart; | |
611 if (residue > occ) { | |
612 /* | |
613 * Residue from previous decode is sufficient | |
614 * to satisfy decode request. Skip to the | |
615 * start of the decoded string, place decoded | |
616 * values in the output buffer, and return. | |
617 */ | |
618 sp->dec_restart += occ; | |
619 do { | |
620 codep = codep->next; | |
621 } while (--residue > occ); | |
622 tp = op + occ; | |
623 do { | |
624 *--tp = codep->value; | |
625 codep = codep->next; | |
626 } while (--occ); | |
627 return (1); | |
628 } | |
629 /* | |
630 * Residue satisfies only part of the decode request. | |
631 */ | |
632 op += residue, occ -= residue; | |
633 tp = op; | |
634 do { | |
635 *--tp = codep->value; | |
636 codep = codep->next; | |
637 } while (--residue); | |
638 sp->dec_restart = 0; | |
639 } | |
640 | |
641 bp = (unsigned char *)tif->tif_rawcp; | |
642 nbits = sp->lzw_nbits; | |
643 nextdata = sp->lzw_nextdata; | |
644 nextbits = sp->lzw_nextbits; | |
645 nbitsmask = sp->dec_nbitsmask; | |
646 oldcodep = sp->dec_oldcodep; | |
647 free_entp = sp->dec_free_entp; | |
648 maxcodep = sp->dec_maxcodep; | |
649 | |
650 while (occ > 0) { | |
651 NextCode(tif, sp, bp, code, GetNextCodeCompat); | |
652 if (code == CODE_EOI) | |
653 break; | |
654 if (code == CODE_CLEAR) { | |
655 free_entp = sp->dec_codetab + CODE_FIRST; | |
656 _TIFFmemset(free_entp, 0, | |
657 (CSIZE - CODE_FIRST) * sizeof (code_t)); | |
658 nbits = BITS_MIN; | |
659 nbitsmask = MAXCODE(BITS_MIN); | |
660 maxcodep = sp->dec_codetab + nbitsmask; | |
661 NextCode(tif, sp, bp, code, GetNextCodeCompat); | |
662 if (code == CODE_EOI) | |
663 break; | |
664 if (code >= CODE_CLEAR) { | |
665 TIFFErrorExt(tif->tif_clientdata, tif->tif_name, | |
666 "LZWDecode: Corrupted LZW table at scanline %d", | |
667 tif->tif_row); | |
668 return (0); | |
669 } | |
670 *op++ = code, occ--; | |
671 oldcodep = sp->dec_codetab + code; | |
672 continue; | |
673 } | |
674 codep = sp->dec_codetab + code; | |
675 | |
676 /* | |
677 * Add the new entry to the code table. | |
678 */ | |
679 if (free_entp < &sp->dec_codetab[0] || | |
680 free_entp >= &sp->dec_codetab[CSIZE]) { | |
681 TIFFErrorExt(tif->tif_clientdata, module, | |
682 "Corrupted LZW table at scanline %d", tif->tif_row); | |
683 return (0); | |
684 } | |
685 | |
686 free_entp->next = oldcodep; | |
687 if (free_entp->next < &sp->dec_codetab[0] || | |
688 free_entp->next >= &sp->dec_codetab[CSIZE]) { | |
689 TIFFErrorExt(tif->tif_clientdata, module, | |
690 "Corrupted LZW table at scanline %d", tif->tif_row); | |
691 return (0); | |
692 } | |
693 free_entp->firstchar = free_entp->next->firstchar; | |
694 free_entp->length = free_entp->next->length+1; | |
695 free_entp->value = (codep < free_entp) ? | |
696 codep->firstchar : free_entp->firstchar; | |
697 if (++free_entp > maxcodep) { | |
698 if (++nbits > BITS_MAX) /* should not happen */ | |
699 nbits = BITS_MAX; | |
700 nbitsmask = MAXCODE(nbits); | |
701 maxcodep = sp->dec_codetab + nbitsmask; | |
702 } | |
703 oldcodep = codep; | |
704 if (code >= 256) { | |
705 /* | |
706 * Code maps to a string, copy string | |
707 * value to output (written in reverse). | |
708 */ | |
709 if(codep->length == 0) { | |
710 TIFFErrorExt(tif->tif_clientdata, module, | |
711 "Wrong length of decoded " | |
712 "string: data probably corrupted at scanline
%d", | |
713 tif->tif_row); | |
714 return (0); | |
715 } | |
716 if (codep->length > occ) { | |
717 /* | |
718 * String is too long for decode buffer, | |
719 * locate portion that will fit, copy to | |
720 * the decode buffer, and setup restart | |
721 * logic for the next decoding call. | |
722 */ | |
723 sp->dec_codep = codep; | |
724 do { | |
725 codep = codep->next; | |
726 } while (codep->length > occ); | |
727 sp->dec_restart = occ; | |
728 tp = op + occ; | |
729 do { | |
730 *--tp = codep->value; | |
731 codep = codep->next; | |
732 } while (--occ); | |
733 break; | |
734 } | |
735 assert(occ >= codep->length); | |
736 op += codep->length, occ -= codep->length; | |
737 tp = op; | |
738 do { | |
739 *--tp = codep->value; | |
740 } while( (codep = codep->next) != NULL ); | |
741 } else | |
742 *op++ = code, occ--; | |
743 } | |
744 | |
745 tif->tif_rawcp = (uint8*) bp; | |
746 sp->lzw_nbits = nbits; | |
747 sp->lzw_nextdata = nextdata; | |
748 sp->lzw_nextbits = nextbits; | |
749 sp->dec_nbitsmask = nbitsmask; | |
750 sp->dec_oldcodep = oldcodep; | |
751 sp->dec_free_entp = free_entp; | |
752 sp->dec_maxcodep = maxcodep; | |
753 | |
754 if (occ > 0) { | |
755 #if defined(__WIN32__) && (defined(_MSC_VER) || defined(__MINGW32__)) | |
756 TIFFErrorExt(tif->tif_clientdata, module, | |
757 "Not enough data at scanline %d (short %I64d bytes)", | |
758 tif->tif_row, (unsigned __int64) occ); | |
759 #else | |
760 TIFFErrorExt(tif->tif_clientdata, module, | |
761 "Not enough data at scanline %d (short %llu bytes)", | |
762 tif->tif_row, (unsigned long long) occ); | |
763 #endif | |
764 return (0); | |
765 } | |
766 return (1); | |
767 } | |
768 #endif /* LZW_COMPAT */ | |
769 | |
770 /* | |
771 * LZW Encoding. | |
772 */ | |
773 | |
774 static int | |
775 LZWSetupEncode(TIFF* tif) | |
776 { | |
777 static const char module[] = "LZWSetupEncode"; | |
778 LZWCodecState* sp = EncoderState(tif); | |
779 | |
780 assert(sp != NULL); | |
781 sp->enc_hashtab = (hash_t*) _TIFFmalloc(HSIZE*sizeof (hash_t)); | |
782 if (sp->enc_hashtab == NULL) { | |
783 TIFFErrorExt(tif->tif_clientdata, module, | |
784 "No space for LZW hash table"); | |
785 return (0); | |
786 } | |
787 return (1); | |
788 } | |
789 | |
790 /* | |
791 * Reset encoding state at the start of a strip. | |
792 */ | |
793 static int | |
794 LZWPreEncode(TIFF* tif, uint16 s) | |
795 { | |
796 LZWCodecState *sp = EncoderState(tif); | |
797 | |
798 (void) s; | |
799 assert(sp != NULL); | |
800 | |
801 if( sp->enc_hashtab == NULL ) | |
802 { | |
803 tif->tif_setupencode( tif ); | |
804 } | |
805 | |
806 sp->lzw_nbits = BITS_MIN; | |
807 sp->lzw_maxcode = MAXCODE(BITS_MIN); | |
808 sp->lzw_free_ent = CODE_FIRST; | |
809 sp->lzw_nextbits = 0; | |
810 sp->lzw_nextdata = 0; | |
811 sp->enc_checkpoint = CHECK_GAP; | |
812 sp->enc_ratio = 0; | |
813 sp->enc_incount = 0; | |
814 sp->enc_outcount = 0; | |
815 /* | |
816 * The 4 here insures there is space for 2 max-sized | |
817 * codes in LZWEncode and LZWPostDecode. | |
818 */ | |
819 sp->enc_rawlimit = tif->tif_rawdata + tif->tif_rawdatasize-1 - 4; | |
820 cl_hash(sp); /* clear hash table */ | |
821 sp->enc_oldcode = (hcode_t) -1; /* generates CODE_CLEAR in LZWEncode */ | |
822 return (1); | |
823 } | |
824 | |
825 #define CALCRATIO(sp, rat) { \ | |
826 if (incount > 0x007fffff) { /* NB: shift will overflow */\ | |
827 rat = outcount >> 8; \ | |
828 rat = (rat == 0 ? 0x7fffffff : incount/rat); \ | |
829 } else \ | |
830 rat = (incount<<8) / outcount; \ | |
831 } | |
832 #define PutNextCode(op, c) { \ | |
833 nextdata = (nextdata << nbits) | c; \ | |
834 nextbits += nbits; \ | |
835 *op++ = (unsigned char)(nextdata >> (nextbits-8)); \ | |
836 nextbits -= 8; \ | |
837 if (nextbits >= 8) { \ | |
838 *op++ = (unsigned char)(nextdata >> (nextbits-8)); \ | |
839 nextbits -= 8; \ | |
840 } \ | |
841 outcount += nbits; \ | |
842 } | |
843 | |
844 /* | |
845 * Encode a chunk of pixels. | |
846 * | |
847 * Uses an open addressing double hashing (no chaining) on the | |
848 * prefix code/next character combination. We do a variant of | |
849 * Knuth's algorithm D (vol. 3, sec. 6.4) along with G. Knott's | |
850 * relatively-prime secondary probe. Here, the modular division | |
851 * first probe is gives way to a faster exclusive-or manipulation. | |
852 * Also do block compression with an adaptive reset, whereby the | |
853 * code table is cleared when the compression ratio decreases, | |
854 * but after the table fills. The variable-length output codes | |
855 * are re-sized at this point, and a CODE_CLEAR is generated | |
856 * for the decoder. | |
857 */ | |
858 static int | |
859 LZWEncode(TIFF* tif, uint8* bp, tmsize_t cc, uint16 s) | |
860 { | |
861 register LZWCodecState *sp = EncoderState(tif); | |
862 register long fcode; | |
863 register hash_t *hp; | |
864 register int h, c; | |
865 hcode_t ent; | |
866 long disp; | |
867 long incount, outcount, checkpoint; | |
868 long nextdata, nextbits; | |
869 int free_ent, maxcode, nbits; | |
870 uint8* op; | |
871 uint8* limit; | |
872 | |
873 (void) s; | |
874 if (sp == NULL) | |
875 return (0); | |
876 | |
877 assert(sp->enc_hashtab != NULL); | |
878 | |
879 /* | |
880 * Load local state. | |
881 */ | |
882 incount = sp->enc_incount; | |
883 outcount = sp->enc_outcount; | |
884 checkpoint = sp->enc_checkpoint; | |
885 nextdata = sp->lzw_nextdata; | |
886 nextbits = sp->lzw_nextbits; | |
887 free_ent = sp->lzw_free_ent; | |
888 maxcode = sp->lzw_maxcode; | |
889 nbits = sp->lzw_nbits; | |
890 op = tif->tif_rawcp; | |
891 limit = sp->enc_rawlimit; | |
892 ent = sp->enc_oldcode; | |
893 | |
894 if (ent == (hcode_t) -1 && cc > 0) { | |
895 /* | |
896 * NB: This is safe because it can only happen | |
897 * at the start of a strip where we know there | |
898 * is space in the data buffer. | |
899 */ | |
900 PutNextCode(op, CODE_CLEAR); | |
901 ent = *bp++; cc--; incount++; | |
902 } | |
903 while (cc > 0) { | |
904 c = *bp++; cc--; incount++; | |
905 fcode = ((long)c << BITS_MAX) + ent; | |
906 h = (c << HSHIFT) ^ ent; /* xor hashing */ | |
907 #ifdef _WINDOWS | |
908 /* | |
909 * Check hash index for an overflow. | |
910 */ | |
911 if (h >= HSIZE) | |
912 h -= HSIZE; | |
913 #endif | |
914 hp = &sp->enc_hashtab[h]; | |
915 if (hp->hash == fcode) { | |
916 ent = hp->code; | |
917 continue; | |
918 } | |
919 if (hp->hash >= 0) { | |
920 /* | |
921 * Primary hash failed, check secondary hash. | |
922 */ | |
923 disp = HSIZE - h; | |
924 if (h == 0) | |
925 disp = 1; | |
926 do { | |
927 /* | |
928 * Avoid pointer arithmetic 'cuz of | |
929 * wraparound problems with segments. | |
930 */ | |
931 if ((h -= disp) < 0) | |
932 h += HSIZE; | |
933 hp = &sp->enc_hashtab[h]; | |
934 if (hp->hash == fcode) { | |
935 ent = hp->code; | |
936 goto hit; | |
937 } | |
938 } while (hp->hash >= 0); | |
939 } | |
940 /* | |
941 * New entry, emit code and add to table. | |
942 */ | |
943 /* | |
944 * Verify there is space in the buffer for the code | |
945 * and any potential Clear code that might be emitted | |
946 * below. The value of limit is setup so that there | |
947 * are at least 4 bytes free--room for 2 codes. | |
948 */ | |
949 if (op > limit) { | |
950 tif->tif_rawcc = (tmsize_t)(op - tif->tif_rawdata); | |
951 TIFFFlushData1(tif); | |
952 op = tif->tif_rawdata; | |
953 } | |
954 PutNextCode(op, ent); | |
955 ent = c; | |
956 hp->code = free_ent++; | |
957 hp->hash = fcode; | |
958 if (free_ent == CODE_MAX-1) { | |
959 /* table is full, emit clear code and reset */ | |
960 cl_hash(sp); | |
961 sp->enc_ratio = 0; | |
962 incount = 0; | |
963 outcount = 0; | |
964 free_ent = CODE_FIRST; | |
965 PutNextCode(op, CODE_CLEAR); | |
966 nbits = BITS_MIN; | |
967 maxcode = MAXCODE(BITS_MIN); | |
968 } else { | |
969 /* | |
970 * If the next entry is going to be too big for | |
971 * the code size, then increase it, if possible. | |
972 */ | |
973 if (free_ent > maxcode) { | |
974 nbits++; | |
975 assert(nbits <= BITS_MAX); | |
976 maxcode = (int) MAXCODE(nbits); | |
977 } else if (incount >= checkpoint) { | |
978 long rat; | |
979 /* | |
980 * Check compression ratio and, if things seem | |
981 * to be slipping, clear the hash table and | |
982 * reset state. The compression ratio is a | |
983 * 24+8-bit fractional number. | |
984 */ | |
985 checkpoint = incount+CHECK_GAP; | |
986 CALCRATIO(sp, rat); | |
987 if (rat <= sp->enc_ratio) { | |
988 cl_hash(sp); | |
989 sp->enc_ratio = 0; | |
990 incount = 0; | |
991 outcount = 0; | |
992 free_ent = CODE_FIRST; | |
993 PutNextCode(op, CODE_CLEAR); | |
994 nbits = BITS_MIN; | |
995 maxcode = MAXCODE(BITS_MIN); | |
996 } else | |
997 sp->enc_ratio = rat; | |
998 } | |
999 } | |
1000 hit: | |
1001 ; | |
1002 } | |
1003 | |
1004 /* | |
1005 * Restore global state. | |
1006 */ | |
1007 sp->enc_incount = incount; | |
1008 sp->enc_outcount = outcount; | |
1009 sp->enc_checkpoint = checkpoint; | |
1010 sp->enc_oldcode = ent; | |
1011 sp->lzw_nextdata = nextdata; | |
1012 sp->lzw_nextbits = nextbits; | |
1013 sp->lzw_free_ent = free_ent; | |
1014 sp->lzw_maxcode = maxcode; | |
1015 sp->lzw_nbits = nbits; | |
1016 tif->tif_rawcp = op; | |
1017 return (1); | |
1018 } | |
1019 | |
1020 /* | |
1021 * Finish off an encoded strip by flushing the last | |
1022 * string and tacking on an End Of Information code. | |
1023 */ | |
1024 static int | |
1025 LZWPostEncode(TIFF* tif) | |
1026 { | |
1027 register LZWCodecState *sp = EncoderState(tif); | |
1028 uint8* op = tif->tif_rawcp; | |
1029 long nextbits = sp->lzw_nextbits; | |
1030 long nextdata = sp->lzw_nextdata; | |
1031 long outcount = sp->enc_outcount; | |
1032 int nbits = sp->lzw_nbits; | |
1033 | |
1034 if (op > sp->enc_rawlimit) { | |
1035 tif->tif_rawcc = (tmsize_t)(op - tif->tif_rawdata); | |
1036 TIFFFlushData1(tif); | |
1037 op = tif->tif_rawdata; | |
1038 } | |
1039 if (sp->enc_oldcode != (hcode_t) -1) { | |
1040 PutNextCode(op, sp->enc_oldcode); | |
1041 sp->enc_oldcode = (hcode_t) -1; | |
1042 } | |
1043 PutNextCode(op, CODE_EOI); | |
1044 if (nextbits > 0) | |
1045 *op++ = (unsigned char)(nextdata << (8-nextbits)); | |
1046 tif->tif_rawcc = (tmsize_t)(op - tif->tif_rawdata); | |
1047 return (1); | |
1048 } | |
1049 | |
1050 /* | |
1051 * Reset encoding hash table. | |
1052 */ | |
1053 static void | |
1054 cl_hash(LZWCodecState* sp) | |
1055 { | |
1056 register hash_t *hp = &sp->enc_hashtab[HSIZE-1]; | |
1057 register long i = HSIZE-8; | |
1058 | |
1059 do { | |
1060 i -= 8; | |
1061 hp[-7].hash = -1; | |
1062 hp[-6].hash = -1; | |
1063 hp[-5].hash = -1; | |
1064 hp[-4].hash = -1; | |
1065 hp[-3].hash = -1; | |
1066 hp[-2].hash = -1; | |
1067 hp[-1].hash = -1; | |
1068 hp[ 0].hash = -1; | |
1069 hp -= 8; | |
1070 } while (i >= 0); | |
1071 for (i += 8; i > 0; i--, hp--) | |
1072 hp->hash = -1; | |
1073 } | |
1074 | |
1075 static void | |
1076 LZWCleanup(TIFF* tif) | |
1077 { | |
1078 (void)TIFFPredictorCleanup(tif); | |
1079 | |
1080 assert(tif->tif_data != 0); | |
1081 | |
1082 if (DecoderState(tif)->dec_codetab) | |
1083 _TIFFfree(DecoderState(tif)->dec_codetab); | |
1084 | |
1085 if (EncoderState(tif)->enc_hashtab) | |
1086 _TIFFfree(EncoderState(tif)->enc_hashtab); | |
1087 | |
1088 _TIFFfree(tif->tif_data); | |
1089 tif->tif_data = NULL; | |
1090 | |
1091 _TIFFSetDefaultCompressionState(tif); | |
1092 } | |
1093 | |
1094 int | |
1095 TIFFInitLZW(TIFF* tif, int scheme) | |
1096 { | |
1097 static const char module[] = "TIFFInitLZW"; | |
1098 assert(scheme == COMPRESSION_LZW); | |
1099 /* | |
1100 * Allocate state block so tag methods have storage to record values. | |
1101 */ | |
1102 tif->tif_data = (uint8*) _TIFFmalloc(sizeof (LZWCodecState)); | |
1103 if (tif->tif_data == NULL) | |
1104 goto bad; | |
1105 DecoderState(tif)->dec_codetab = NULL; | |
1106 DecoderState(tif)->dec_decode = NULL; | |
1107 EncoderState(tif)->enc_hashtab = NULL; | |
1108 LZWState(tif)->rw_mode = tif->tif_mode; | |
1109 | |
1110 /* | |
1111 * Install codec methods. | |
1112 */ | |
1113 tif->tif_fixuptags = LZWFixupTags; | |
1114 tif->tif_setupdecode = LZWSetupDecode; | |
1115 tif->tif_predecode = LZWPreDecode; | |
1116 tif->tif_decoderow = LZWDecode; | |
1117 tif->tif_decodestrip = LZWDecode; | |
1118 tif->tif_decodetile = LZWDecode; | |
1119 tif->tif_setupencode = LZWSetupEncode; | |
1120 tif->tif_preencode = LZWPreEncode; | |
1121 tif->tif_postencode = LZWPostEncode; | |
1122 tif->tif_encoderow = LZWEncode; | |
1123 tif->tif_encodestrip = LZWEncode; | |
1124 tif->tif_encodetile = LZWEncode; | |
1125 tif->tif_cleanup = LZWCleanup; | |
1126 /* | |
1127 * Setup predictor setup. | |
1128 */ | |
1129 (void) TIFFPredictorInit(tif); | |
1130 return (1); | |
1131 bad: | |
1132 TIFFErrorExt(tif->tif_clientdata, module, | |
1133 "No space for LZW state block"); | |
1134 return (0); | |
1135 } | |
1136 | |
1137 /* | |
1138 * Copyright (c) 1985, 1986 The Regents of the University of California. | |
1139 * All rights reserved. | |
1140 * | |
1141 * This code is derived from software contributed to Berkeley by | |
1142 * James A. Woods, derived from original work by Spencer Thomas | |
1143 * and Joseph Orost. | |
1144 * | |
1145 * Redistribution and use in source and binary forms are permitted | |
1146 * provided that the above copyright notice and this paragraph are | |
1147 * duplicated in all such forms and that any documentation, | |
1148 * advertising materials, and other materials related to such | |
1149 * distribution and use acknowledge that the software was developed | |
1150 * by the University of California, Berkeley. The name of the | |
1151 * University may not be used to endorse or promote products derived | |
1152 * from this software without specific prior written permission. | |
1153 * THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR | |
1154 * IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED | |
1155 * WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR A PARTICULAR PURPOSE. | |
1156 */ | |
1157 #endif /* LZW_SUPPORT */ | |
1158 | |
1159 /* vim: set ts=8 sts=8 sw=8 noet: */ | |
1160 /* | |
1161 * Local Variables: | |
1162 * mode: c | |
1163 * c-basic-offset: 8 | |
1164 * fill-column: 78 | |
1165 * End: | |
1166 */ | |
1167 | |
OLD | NEW |