1 |
/* inflate.c -- zlib decompression |
2 |
* Copyright (C) 1995-2010 Mark Adler |
3 |
* For conditions of distribution and use, see copyright notice in zlib.h |
4 |
*/ |
5 |
|
6 |
/* |
7 |
* Change history: |
8 |
* |
9 |
* 1.2.beta0 24 Nov 2002 |
10 |
* - First version -- complete rewrite of inflate to simplify code, avoid |
11 |
* creation of window when not needed, minimize use of window when it is |
12 |
* needed, make inffast.c even faster, implement gzip decoding, and to |
13 |
* improve code readability and style over the previous zlib inflate code |
14 |
* |
15 |
* 1.2.beta1 25 Nov 2002 |
16 |
* - Use pointers for available input and output checking in inffast.c |
17 |
* - Remove input and output counters in inffast.c |
18 |
* - Change inffast.c entry and loop from avail_in >= 7 to >= 6 |
19 |
* - Remove unnecessary second byte pull from length extra in inffast.c |
20 |
* - Unroll direct copy to three copies per loop in inffast.c |
21 |
* |
22 |
* 1.2.beta2 4 Dec 2002 |
23 |
* - Change external routine names to reduce potential conflicts |
24 |
* - Correct filename to inffixed.h for fixed tables in inflate.c |
25 |
* - Make hbuf[] unsigned char to match parameter type in inflate.c |
26 |
* - Change strm->next_out[-state->offset] to *(strm->next_out - state->offset) |
27 |
* to avoid negation problem on Alphas (64 bit) in inflate.c |
28 |
* |
29 |
* 1.2.beta3 22 Dec 2002 |
30 |
* - Add comments on state->bits assertion in inffast.c |
31 |
* - Add comments on op field in inftrees.h |
32 |
* - Fix bug in reuse of allocated window after inflateReset() |
33 |
* - Remove bit fields--back to byte structure for speed |
34 |
* - Remove distance extra == 0 check in inflate_fast()--only helps for lengths |
35 |
* - Change post-increments to pre-increments in inflate_fast(), PPC biased? |
36 |
* - Add compile time option, POSTINC, to use post-increments instead (Intel?) |
37 |
* - Make MATCH copy in inflate() much faster for when inflate_fast() not used |
38 |
* - Use local copies of stream next and avail values, as well as local bit |
39 |
* buffer and bit count in inflate()--for speed when inflate_fast() not used |
40 |
* |
41 |
* 1.2.beta4 1 Jan 2003 |
42 |
* - Split ptr - 257 statements in inflate_table() to avoid compiler warnings |
43 |
* - Move a comment on output buffer sizes from inffast.c to inflate.c |
44 |
* - Add comments in inffast.c to introduce the inflate_fast() routine |
45 |
* - Rearrange window copies in inflate_fast() for speed and simplification |
46 |
* - Unroll last copy for window match in inflate_fast() |
47 |
* - Use local copies of window variables in inflate_fast() for speed |
48 |
* - Pull out common wnext == 0 case for speed in inflate_fast() |
49 |
* - Make op and len in inflate_fast() unsigned for consistency |
50 |
* - Add FAR to lcode and dcode declarations in inflate_fast() |
51 |
* - Simplified bad distance check in inflate_fast() |
52 |
* - Added inflateBackInit(), inflateBack(), and inflateBackEnd() in new |
53 |
* source file infback.c to provide a call-back interface to inflate for |
54 |
* programs like gzip and unzip -- uses window as output buffer to avoid |
55 |
* window copying |
56 |
* |
57 |
* 1.2.beta5 1 Jan 2003 |
58 |
* - Improved inflateBack() interface to allow the caller to provide initial |
59 |
* input in strm. |
60 |
* - Fixed stored blocks bug in inflateBack() |
61 |
* |
62 |
* 1.2.beta6 4 Jan 2003 |
63 |
* - Added comments in inffast.c on effectiveness of POSTINC |
64 |
* - Typecasting all around to reduce compiler warnings |
65 |
* - Changed loops from while (1) or do {} while (1) to for (;;), again to |
66 |
* make compilers happy |
67 |
* - Changed type of window in inflateBackInit() to unsigned char * |
68 |
* |
69 |
* 1.2.beta7 27 Jan 2003 |
70 |
* - Changed many types to unsigned or unsigned short to avoid warnings |
71 |
* - Added inflateCopy() function |
72 |
* |
73 |
* 1.2.0 9 Mar 2003 |
74 |
* - Changed inflateBack() interface to provide separate opaque descriptors |
75 |
* for the in() and out() functions |
76 |
* - Changed inflateBack() argument and in_func typedef to swap the length |
77 |
* and buffer address return values for the input function |
78 |
* - Check next_in and next_out for Z_NULL on entry to inflate() |
79 |
* |
80 |
* The history for versions after 1.2.0 are in ChangeLog in zlib distribution. |
81 |
*/ |
82 |
|
83 |
#include "zutil.h" |
84 |
#include "inftrees.h" |
85 |
#include "inflate.h" |
86 |
#include "inffast.h" |
87 |
|
88 |
#ifdef MAKEFIXED |
89 |
# ifndef BUILDFIXED |
90 |
# define BUILDFIXED |
91 |
# endif |
92 |
#endif |
93 |
|
94 |
/* function prototypes */ |
95 |
local void fixedtables OF((struct inflate_state FAR *state)); |
96 |
local int updatewindow OF((z_streamp strm, unsigned out)); |
97 |
#ifdef BUILDFIXED |
98 |
void makefixed OF((void)); |
99 |
#endif |
100 |
local unsigned syncsearch OF((unsigned FAR *have, unsigned char FAR *buf, |
101 |
unsigned len)); |
102 |
|
103 |
int ZEXPORT inflateReset(strm) |
104 |
z_streamp strm; |
105 |
{ |
106 |
struct inflate_state FAR *state; |
107 |
|
108 |
if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; |
109 |
state = (struct inflate_state FAR *)strm->state; |
110 |
strm->total_in = strm->total_out = state->total = 0; |
111 |
strm->msg = Z_NULL; |
112 |
strm->adler = 1; /* to support ill-conceived Java test suite */ |
113 |
state->mode = HEAD; |
114 |
state->last = 0; |
115 |
state->havedict = 0; |
116 |
state->dmax = 32768U; |
117 |
state->head = Z_NULL; |
118 |
state->wsize = 0; |
119 |
state->whave = 0; |
120 |
state->wnext = 0; |
121 |
state->hold = 0; |
122 |
state->bits = 0; |
123 |
state->lencode = state->distcode = state->next = state->codes; |
124 |
state->sane = 1; |
125 |
state->back = -1; |
126 |
Tracev((stderr, "inflate: reset\n")); |
127 |
return Z_OK; |
128 |
} |
129 |
|
130 |
int ZEXPORT inflateReset2(strm, windowBits) |
131 |
z_streamp strm; |
132 |
int windowBits; |
133 |
{ |
134 |
int wrap; |
135 |
struct inflate_state FAR *state; |
136 |
|
137 |
/* get the state */ |
138 |
if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; |
139 |
state = (struct inflate_state FAR *)strm->state; |
140 |
|
141 |
/* extract wrap request from windowBits parameter */ |
142 |
if (windowBits < 0) { |
143 |
wrap = 0; |
144 |
windowBits = -windowBits; |
145 |
} |
146 |
else { |
147 |
wrap = (windowBits >> 4) + 1; |
148 |
#ifdef GUNZIP |
149 |
if (windowBits < 48) |
150 |
windowBits &= 15; |
151 |
#endif |
152 |
} |
153 |
|
154 |
/* set number of window bits, free window if different */ |
155 |
if (windowBits && (windowBits < 8 || windowBits > 15)) |
156 |
return Z_STREAM_ERROR; |
157 |
if (state->window != Z_NULL && state->wbits != (unsigned)windowBits) { |
158 |
ZFREE(strm, state->window); |
159 |
state->window = Z_NULL; |
160 |
} |
161 |
|
162 |
/* update state and reset the rest of it */ |
163 |
state->wrap = wrap; |
164 |
state->wbits = (unsigned)windowBits; |
165 |
return inflateReset(strm); |
166 |
} |
167 |
|
168 |
int ZEXPORT inflateInit2_(strm, windowBits, version, stream_size) |
169 |
z_streamp strm; |
170 |
int windowBits; |
171 |
const char *version; |
172 |
int stream_size; |
173 |
{ |
174 |
int ret; |
175 |
struct inflate_state FAR *state; |
176 |
|
177 |
if (version == Z_NULL || version[0] != ZLIB_VERSION[0] || |
178 |
stream_size != (int)(sizeof(z_stream))) |
179 |
return Z_VERSION_ERROR; |
180 |
if (strm == Z_NULL) return Z_STREAM_ERROR; |
181 |
strm->msg = Z_NULL; /* in case we return an error */ |
182 |
if (strm->zalloc == (alloc_func)0) { |
183 |
strm->zalloc = zcalloc; |
184 |
strm->opaque = (voidpf)0; |
185 |
} |
186 |
if (strm->zfree == (free_func)0) strm->zfree = zcfree; |
187 |
state = (struct inflate_state FAR *) |
188 |
ZALLOC(strm, 1, sizeof(struct inflate_state)); |
189 |
if (state == Z_NULL) return Z_MEM_ERROR; |
190 |
Tracev((stderr, "inflate: allocated\n")); |
191 |
strm->state = (struct internal_state FAR *)state; |
192 |
state->window = Z_NULL; |
193 |
ret = inflateReset2(strm, windowBits); |
194 |
if (ret != Z_OK) { |
195 |
ZFREE(strm, state); |
196 |
strm->state = Z_NULL; |
197 |
} |
198 |
return ret; |
199 |
} |
200 |
|
201 |
int ZEXPORT inflateInit_(strm, version, stream_size) |
202 |
z_streamp strm; |
203 |
const char *version; |
204 |
int stream_size; |
205 |
{ |
206 |
return inflateInit2_(strm, DEF_WBITS, version, stream_size); |
207 |
} |
208 |
|
209 |
int ZEXPORT inflatePrime(strm, bits, value) |
210 |
z_streamp strm; |
211 |
int bits; |
212 |
int value; |
213 |
{ |
214 |
struct inflate_state FAR *state; |
215 |
|
216 |
if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; |
217 |
state = (struct inflate_state FAR *)strm->state; |
218 |
if (bits < 0) { |
219 |
state->hold = 0; |
220 |
state->bits = 0; |
221 |
return Z_OK; |
222 |
} |
223 |
if (bits > 16 || state->bits + bits > 32) return Z_STREAM_ERROR; |
224 |
value &= (1L << bits) - 1; |
225 |
state->hold += value << state->bits; |
226 |
state->bits += bits; |
227 |
return Z_OK; |
228 |
} |
229 |
|
230 |
/* |
231 |
Return state with length and distance decoding tables and index sizes set to |
232 |
fixed code decoding. Normally this returns fixed tables from inffixed.h. |
233 |
If BUILDFIXED is defined, then instead this routine builds the tables the |
234 |
first time it's called, and returns those tables the first time and |
235 |
thereafter. This reduces the size of the code by about 2K bytes, in |
236 |
exchange for a little execution time. However, BUILDFIXED should not be |
237 |
used for threaded applications, since the rewriting of the tables and virgin |
238 |
may not be thread-safe. |
239 |
*/ |
240 |
local void fixedtables(state) |
241 |
struct inflate_state FAR *state; |
242 |
{ |
243 |
#ifdef BUILDFIXED |
244 |
static int virgin = 1; |
245 |
static code *lenfix, *distfix; |
246 |
static code fixed[544]; |
247 |
|
248 |
/* build fixed huffman tables if first call (may not be thread safe) */ |
249 |
if (virgin) { |
250 |
unsigned sym, bits; |
251 |
static code *next; |
252 |
|
253 |
/* literal/length table */ |
254 |
sym = 0; |
255 |
while (sym < 144) state->lens[sym++] = 8; |
256 |
while (sym < 256) state->lens[sym++] = 9; |
257 |
while (sym < 280) state->lens[sym++] = 7; |
258 |
while (sym < 288) state->lens[sym++] = 8; |
259 |
next = fixed; |
260 |
lenfix = next; |
261 |
bits = 9; |
262 |
inflate_table(LENS, state->lens, 288, &(next), &(bits), state->work); |
263 |
|
264 |
/* distance table */ |
265 |
sym = 0; |
266 |
while (sym < 32) state->lens[sym++] = 5; |
267 |
distfix = next; |
268 |
bits = 5; |
269 |
inflate_table(DISTS, state->lens, 32, &(next), &(bits), state->work); |
270 |
|
271 |
/* do this just once */ |
272 |
virgin = 0; |
273 |
} |
274 |
#else /* !BUILDFIXED */ |
275 |
# include "inffixed.h" |
276 |
#endif /* BUILDFIXED */ |
277 |
state->lencode = lenfix; |
278 |
state->lenbits = 9; |
279 |
state->distcode = distfix; |
280 |
state->distbits = 5; |
281 |
} |
282 |
|
283 |
#ifdef MAKEFIXED |
284 |
#include <stdio.h> |
285 |
|
286 |
/* |
287 |
Write out the inffixed.h that is #include'd above. Defining MAKEFIXED also |
288 |
defines BUILDFIXED, so the tables are built on the fly. makefixed() writes |
289 |
those tables to stdout, which would be piped to inffixed.h. A small program |
290 |
can simply call makefixed to do this: |
291 |
|
292 |
void makefixed(void); |
293 |
|
294 |
int main(void) |
295 |
{ |
296 |
makefixed(); |
297 |
return 0; |
298 |
} |
299 |
|
300 |
Then that can be linked with zlib built with MAKEFIXED defined and run: |
301 |
|
302 |
a.out > inffixed.h |
303 |
*/ |
304 |
void makefixed() |
305 |
{ |
306 |
unsigned low, size; |
307 |
struct inflate_state state; |
308 |
|
309 |
fixedtables(&state); |
310 |
puts(" /* inffixed.h -- table for decoding fixed codes"); |
311 |
puts(" * Generated automatically by makefixed()."); |
312 |
puts(" */"); |
313 |
puts(""); |
314 |
puts(" /* WARNING: this file should *not* be used by applications."); |
315 |
puts(" It is part of the implementation of this library and is"); |
316 |
puts(" subject to change. Applications should only use zlib.h."); |
317 |
puts(" */"); |
318 |
puts(""); |
319 |
size = 1U << 9; |
320 |
printf(" static const code lenfix[%u] = {", size); |
321 |
low = 0; |
322 |
for (;;) { |
323 |
if ((low % 7) == 0) printf("\n "); |
324 |
printf("{%u,%u,%d}", state.lencode[low].op, state.lencode[low].bits, |
325 |
state.lencode[low].val); |
326 |
if (++low == size) break; |
327 |
putchar(','); |
328 |
} |
329 |
puts("\n };"); |
330 |
size = 1U << 5; |
331 |
printf("\n static const code distfix[%u] = {", size); |
332 |
low = 0; |
333 |
for (;;) { |
334 |
if ((low % 6) == 0) printf("\n "); |
335 |
printf("{%u,%u,%d}", state.distcode[low].op, state.distcode[low].bits, |
336 |
state.distcode[low].val); |
337 |
if (++low == size) break; |
338 |
putchar(','); |
339 |
} |
340 |
puts("\n };"); |
341 |
} |
342 |
#endif /* MAKEFIXED */ |
343 |
|
344 |
/* |
345 |
Update the window with the last wsize (normally 32K) bytes written before |
346 |
returning. If window does not exist yet, create it. This is only called |
347 |
when a window is already in use, or when output has been written during this |
348 |
inflate call, but the end of the deflate stream has not been reached yet. |
349 |
It is also called to create a window for dictionary data when a dictionary |
350 |
is loaded. |
351 |
|
352 |
Providing output buffers larger than 32K to inflate() should provide a speed |
353 |
advantage, since only the last 32K of output is copied to the sliding window |
354 |
upon return from inflate(), and since all distances after the first 32K of |
355 |
output will fall in the output data, making match copies simpler and faster. |
356 |
The advantage may be dependent on the size of the processor's data caches. |
357 |
*/ |
358 |
local int updatewindow(strm, out) |
359 |
z_streamp strm; |
360 |
unsigned out; |
361 |
{ |
362 |
struct inflate_state FAR *state; |
363 |
unsigned copy, dist; |
364 |
|
365 |
state = (struct inflate_state FAR *)strm->state; |
366 |
|
367 |
/* if it hasn't been done already, allocate space for the window */ |
368 |
if (state->window == Z_NULL) { |
369 |
state->window = (unsigned char FAR *) |
370 |
ZALLOC(strm, 1U << state->wbits, |
371 |
sizeof(unsigned char)); |
372 |
if (state->window == Z_NULL) return 1; |
373 |
} |
374 |
|
375 |
/* if window not in use yet, initialize */ |
376 |
if (state->wsize == 0) { |
377 |
state->wsize = 1U << state->wbits; |
378 |
state->wnext = 0; |
379 |
state->whave = 0; |
380 |
} |
381 |
|
382 |
/* copy state->wsize or less output bytes into the circular window */ |
383 |
copy = out - strm->avail_out; |
384 |
if (copy >= state->wsize) { |
385 |
zmemcpy(state->window, strm->next_out - state->wsize, state->wsize); |
386 |
state->wnext = 0; |
387 |
state->whave = state->wsize; |
388 |
} |
389 |
else { |
390 |
dist = state->wsize - state->wnext; |
391 |
if (dist > copy) dist = copy; |
392 |
zmemcpy(state->window + state->wnext, strm->next_out - copy, dist); |
393 |
copy -= dist; |
394 |
if (copy) { |
395 |
zmemcpy(state->window, strm->next_out - copy, copy); |
396 |
state->wnext = copy; |
397 |
state->whave = state->wsize; |
398 |
} |
399 |
else { |
400 |
state->wnext += dist; |
401 |
if (state->wnext == state->wsize) state->wnext = 0; |
402 |
if (state->whave < state->wsize) state->whave += dist; |
403 |
} |
404 |
} |
405 |
return 0; |
406 |
} |
407 |
|
408 |
/* Macros for inflate(): */ |
409 |
|
410 |
/* check function to use adler32() for zlib or crc32() for gzip */ |
411 |
#ifdef GUNZIP |
412 |
# define UPDATE(check, buf, len) \ |
413 |
(state->flags ? crc32(check, buf, len) : adler32(check, buf, len)) |
414 |
#else |
415 |
# define UPDATE(check, buf, len) adler32(check, buf, len) |
416 |
#endif |
417 |
|
418 |
/* check macros for header crc */ |
419 |
#ifdef GUNZIP |
420 |
# define CRC2(check, word) \ |
421 |
do { \ |
422 |
hbuf[0] = (unsigned char)(word); \ |
423 |
hbuf[1] = (unsigned char)((word) >> 8); \ |
424 |
check = crc32(check, hbuf, 2); \ |
425 |
} while (0) |
426 |
|
427 |
# define CRC4(check, word) \ |
428 |
do { \ |
429 |
hbuf[0] = (unsigned char)(word); \ |
430 |
hbuf[1] = (unsigned char)((word) >> 8); \ |
431 |
hbuf[2] = (unsigned char)((word) >> 16); \ |
432 |
hbuf[3] = (unsigned char)((word) >> 24); \ |
433 |
check = crc32(check, hbuf, 4); \ |
434 |
} while (0) |
435 |
#endif |
436 |
|
437 |
/* Load registers with state in inflate() for speed */ |
438 |
#define LOAD() \ |
439 |
do { \ |
440 |
put = strm->next_out; \ |
441 |
left = strm->avail_out; \ |
442 |
next = strm->next_in; \ |
443 |
have = strm->avail_in; \ |
444 |
hold = state->hold; \ |
445 |
bits = state->bits; \ |
446 |
} while (0) |
447 |
|
448 |
/* Restore state from registers in inflate() */ |
449 |
#define RESTORE() \ |
450 |
do { \ |
451 |
strm->next_out = put; \ |
452 |
strm->avail_out = left; \ |
453 |
strm->next_in = next; \ |
454 |
strm->avail_in = have; \ |
455 |
state->hold = hold; \ |
456 |
state->bits = bits; \ |
457 |
} while (0) |
458 |
|
459 |
/* Clear the input bit accumulator */ |
460 |
#define INITBITS() \ |
461 |
do { \ |
462 |
hold = 0; \ |
463 |
bits = 0; \ |
464 |
} while (0) |
465 |
|
466 |
/* Get a byte of input into the bit accumulator, or return from inflate() |
467 |
if there is no input available. */ |
468 |
#define PULLBYTE() \ |
469 |
do { \ |
470 |
if (have == 0) goto inf_leave; \ |
471 |
have--; \ |
472 |
hold += (unsigned long)(*next++) << bits; \ |
473 |
bits += 8; \ |
474 |
} while (0) |
475 |
|
476 |
/* Assure that there are at least n bits in the bit accumulator. If there is |
477 |
not enough available input to do that, then return from inflate(). */ |
478 |
#define NEEDBITS(n) \ |
479 |
do { \ |
480 |
while (bits < (unsigned)(n)) \ |
481 |
PULLBYTE(); \ |
482 |
} while (0) |
483 |
|
484 |
/* Return the low n bits of the bit accumulator (n < 16) */ |
485 |
#define BITS(n) \ |
486 |
((unsigned)hold & ((1U << (n)) - 1)) |
487 |
|
488 |
/* Remove n bits from the bit accumulator */ |
489 |
#define DROPBITS(n) \ |
490 |
do { \ |
491 |
hold >>= (n); \ |
492 |
bits -= (unsigned)(n); \ |
493 |
} while (0) |
494 |
|
495 |
/* Remove zero to seven bits as needed to go to a byte boundary */ |
496 |
#define BYTEBITS() \ |
497 |
do { \ |
498 |
hold >>= bits & 7; \ |
499 |
bits -= bits & 7; \ |
500 |
} while (0) |
501 |
|
502 |
/* Reverse the bytes in a 32-bit value */ |
503 |
#define REVERSE(q) \ |
504 |
((((q) >> 24) & 0xff) + (((q) >> 8) & 0xff00) + \ |
505 |
(((q) & 0xff00) << 8) + (((q) & 0xff) << 24)) |
506 |
|
507 |
/* |
508 |
inflate() uses a state machine to process as much input data and generate as |
509 |
much output data as possible before returning. The state machine is |
510 |
structured roughly as follows: |
511 |
|
512 |
for (;;) switch (state) { |
513 |
... |
514 |
case STATEn: |
515 |
if (not enough input data or output space to make progress) |
516 |
return; |
517 |
... make progress ... |
518 |
state = STATEm; |
519 |
break; |
520 |
... |
521 |
} |
522 |
|
523 |
so when inflate() is called again, the same case is attempted again, and |
524 |
if the appropriate resources are provided, the machine proceeds to the |
525 |
next state. The NEEDBITS() macro is usually the way the state evaluates |
526 |
whether it can proceed or should return. NEEDBITS() does the return if |
527 |
the requested bits are not available. The typical use of the BITS macros |
528 |
is: |
529 |
|
530 |
NEEDBITS(n); |
531 |
... do something with BITS(n) ... |
532 |
DROPBITS(n); |
533 |
|
534 |
where NEEDBITS(n) either returns from inflate() if there isn't enough |
535 |
input left to load n bits into the accumulator, or it continues. BITS(n) |
536 |
gives the low n bits in the accumulator. When done, DROPBITS(n) drops |
537 |
the low n bits off the accumulator. INITBITS() clears the accumulator |
538 |
and sets the number of available bits to zero. BYTEBITS() discards just |
539 |
enough bits to put the accumulator on a byte boundary. After BYTEBITS() |
540 |
and a NEEDBITS(8), then BITS(8) would return the next byte in the stream. |
541 |
|
542 |
NEEDBITS(n) uses PULLBYTE() to get an available byte of input, or to return |
543 |
if there is no input available. The decoding of variable length codes uses |
544 |
PULLBYTE() directly in order to pull just enough bytes to decode the next |
545 |
code, and no more. |
546 |
|
547 |
Some states loop until they get enough input, making sure that enough |
548 |
state information is maintained to continue the loop where it left off |
549 |
if NEEDBITS() returns in the loop. For example, want, need, and keep |
550 |
would all have to actually be part of the saved state in case NEEDBITS() |
551 |
returns: |
552 |
|
553 |
case STATEw: |
554 |
while (want < need) { |
555 |
NEEDBITS(n); |
556 |
keep[want++] = BITS(n); |
557 |
DROPBITS(n); |
558 |
} |
559 |
state = STATEx; |
560 |
case STATEx: |
561 |
|
562 |
As shown above, if the next state is also the next case, then the break |
563 |
is omitted. |
564 |
|
565 |
A state may also return if there is not enough output space available to |
566 |
complete that state. Those states are copying stored data, writing a |
567 |
literal byte, and copying a matching string. |
568 |
|
569 |
When returning, a "goto inf_leave" is used to update the total counters, |
570 |
update the check value, and determine whether any progress has been made |
571 |
during that inflate() call in order to return the proper return code. |
572 |
Progress is defined as a change in either strm->avail_in or strm->avail_out. |
573 |
When there is a window, goto inf_leave will update the window with the last |
574 |
output written. If a goto inf_leave occurs in the middle of decompression |
575 |
and there is no window currently, goto inf_leave will create one and copy |
576 |
output to the window for the next call of inflate(). |
577 |
|
578 |
In this implementation, the flush parameter of inflate() only affects the |
579 |
return code (per zlib.h). inflate() always writes as much as possible to |
580 |
strm->next_out, given the space available and the provided input--the effect |
581 |
documented in zlib.h of Z_SYNC_FLUSH. Furthermore, inflate() always defers |
582 |
the allocation of and copying into a sliding window until necessary, which |
583 |
provides the effect documented in zlib.h for Z_FINISH when the entire input |
584 |
stream available. So the only thing the flush parameter actually does is: |
585 |
when flush is set to Z_FINISH, inflate() cannot return Z_OK. Instead it |
586 |
will return Z_BUF_ERROR if it has not reached the end of the stream. |
587 |
*/ |
588 |
|
589 |
int ZEXPORT inflate(strm, flush) |
590 |
z_streamp strm; |
591 |
int flush; |
592 |
{ |
593 |
struct inflate_state FAR *state; |
594 |
unsigned char FAR *next; /* next input */ |
595 |
unsigned char FAR *put; /* next output */ |
596 |
unsigned have, left; /* available input and output */ |
597 |
unsigned long hold; /* bit buffer */ |
598 |
unsigned bits; /* bits in bit buffer */ |
599 |
unsigned in, out; /* save starting available input and output */ |
600 |
unsigned copy; /* number of stored or match bytes to copy */ |
601 |
unsigned char FAR *from; /* where to copy match bytes from */ |
602 |
code here; /* current decoding table entry */ |
603 |
code last; /* parent table entry */ |
604 |
unsigned len; /* length to copy for repeats, bits to drop */ |
605 |
int ret; /* return code */ |
606 |
#ifdef GUNZIP |
607 |
unsigned char hbuf[4]; /* buffer for gzip header crc calculation */ |
608 |
#endif |
609 |
static const unsigned short order[19] = /* permutation of code lengths */ |
610 |
{16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15}; |
611 |
|
612 |
if (strm == Z_NULL || strm->state == Z_NULL || strm->next_out == Z_NULL || |
613 |
(strm->next_in == Z_NULL && strm->avail_in != 0)) |
614 |
return Z_STREAM_ERROR; |
615 |
|
616 |
state = (struct inflate_state FAR *)strm->state; |
617 |
if (state->mode == TYPE) state->mode = TYPEDO; /* skip check */ |
618 |
LOAD(); |
619 |
in = have; |
620 |
out = left; |
621 |
ret = Z_OK; |
622 |
for (;;) |
623 |
switch (state->mode) { |
624 |
case HEAD: |
625 |
if (state->wrap == 0) { |
626 |
state->mode = TYPEDO; |
627 |
break; |
628 |
} |
629 |
NEEDBITS(16); |
630 |
#ifdef GUNZIP |
631 |
if ((state->wrap & 2) && hold == 0x8b1f) { /* gzip header */ |
632 |
state->check = crc32(0L, Z_NULL, 0); |
633 |
CRC2(state->check, hold); |
634 |
INITBITS(); |
635 |
state->mode = FLAGS; |
636 |
break; |
637 |
} |
638 |
state->flags = 0; /* expect zlib header */ |
639 |
if (state->head != Z_NULL) |
640 |
state->head->done = -1; |
641 |
if (!(state->wrap & 1) || /* check if zlib header allowed */ |
642 |
#else |
643 |
if ( |
644 |
#endif |
645 |
((BITS(8) << 8) + (hold >> 8)) % 31) { |
646 |
strm->msg = (char *)"incorrect header check"; |
647 |
state->mode = BAD; |
648 |
break; |
649 |
} |
650 |
if (BITS(4) != Z_DEFLATED) { |
651 |
strm->msg = (char *)"unknown compression method"; |
652 |
state->mode = BAD; |
653 |
break; |
654 |
} |
655 |
DROPBITS(4); |
656 |
len = BITS(4) + 8; |
657 |
if (state->wbits == 0) |
658 |
state->wbits = len; |
659 |
else if (len > state->wbits) { |
660 |
strm->msg = (char *)"invalid window size"; |
661 |
state->mode = BAD; |
662 |
break; |
663 |
} |
664 |
state->dmax = 1U << len; |
665 |
Tracev((stderr, "inflate: zlib header ok\n")); |
666 |
strm->adler = state->check = adler32(0L, Z_NULL, 0); |
667 |
state->mode = hold & 0x200 ? DICTID : TYPE; |
668 |
INITBITS(); |
669 |
break; |
670 |
#ifdef GUNZIP |
671 |
case FLAGS: |
672 |
NEEDBITS(16); |
673 |
state->flags = (int)(hold); |
674 |
if ((state->flags & 0xff) != Z_DEFLATED) { |
675 |
strm->msg = (char *)"unknown compression method"; |
676 |
state->mode = BAD; |
677 |
break; |
678 |
} |
679 |
if (state->flags & 0xe000) { |
680 |
strm->msg = (char *)"unknown header flags set"; |
681 |
state->mode = BAD; |
682 |
break; |
683 |
} |
684 |
if (state->head != Z_NULL) |
685 |
state->head->text = (int)((hold >> 8) & 1); |
686 |
if (state->flags & 0x0200) CRC2(state->check, hold); |
687 |
INITBITS(); |
688 |
state->mode = TIME; |
689 |
case TIME: |
690 |
NEEDBITS(32); |
691 |
if (state->head != Z_NULL) |
692 |
state->head->time = hold; |
693 |
if (state->flags & 0x0200) CRC4(state->check, hold); |
694 |
INITBITS(); |
695 |
state->mode = OS; |
696 |
case OS: |
697 |
NEEDBITS(16); |
698 |
if (state->head != Z_NULL) { |
699 |
state->head->xflags = (int)(hold & 0xff); |
700 |
state->head->os = (int)(hold >> 8); |
701 |
} |
702 |
if (state->flags & 0x0200) CRC2(state->check, hold); |
703 |
INITBITS(); |
704 |
state->mode = EXLEN; |
705 |
case EXLEN: |
706 |
if (state->flags & 0x0400) { |
707 |
NEEDBITS(16); |
708 |
state->length = (unsigned)(hold); |
709 |
if (state->head != Z_NULL) |
710 |
state->head->extra_len = (unsigned)hold; |
711 |
if (state->flags & 0x0200) CRC2(state->check, hold); |
712 |
INITBITS(); |
713 |
} |
714 |
else if (state->head != Z_NULL) |
715 |
state->head->extra = Z_NULL; |
716 |
state->mode = EXTRA; |
717 |
case EXTRA: |
718 |
if (state->flags & 0x0400) { |
719 |
copy = state->length; |
720 |
if (copy > have) copy = have; |
721 |
if (copy) { |
722 |
if (state->head != Z_NULL && |
723 |
state->head->extra != Z_NULL) { |
724 |
len = state->head->extra_len - state->length; |
725 |
zmemcpy(state->head->extra + len, next, |
726 |
len + copy > state->head->extra_max ? |
727 |
state->head->extra_max - len : copy); |
728 |
} |
729 |
if (state->flags & 0x0200) |
730 |
state->check = crc32(state->check, next, copy); |
731 |
have -= copy; |
732 |
next += copy; |
733 |
state->length -= copy; |
734 |
} |
735 |
if (state->length) goto inf_leave; |
736 |
} |
737 |
state->length = 0; |
738 |
state->mode = NAME; |
739 |
case NAME: |
740 |
if (state->flags & 0x0800) { |
741 |
if (have == 0) goto inf_leave; |
742 |
copy = 0; |
743 |
do { |
744 |
len = (unsigned)(next[copy++]); |
745 |
if (state->head != Z_NULL && |
746 |
state->head->name != Z_NULL && |
747 |
state->length < state->head->name_max) |
748 |
state->head->name[state->length++] = len; |
749 |
} while (len && copy < have); |
750 |
if (state->flags & 0x0200) |
751 |
state->check = crc32(state->check, next, copy); |
752 |
have -= copy; |
753 |
next += copy; |
754 |
if (len) goto inf_leave; |
755 |
} |
756 |
else if (state->head != Z_NULL) |
757 |
state->head->name = Z_NULL; |
758 |
state->length = 0; |
759 |
state->mode = COMMENT; |
760 |
case COMMENT: |
761 |
if (state->flags & 0x1000) { |
762 |
if (have == 0) goto inf_leave; |
763 |
copy = 0; |
764 |
do { |
765 |
len = (unsigned)(next[copy++]); |
766 |
if (state->head != Z_NULL && |
767 |
state->head->comment != Z_NULL && |
768 |
state->length < state->head->comm_max) |
769 |
state->head->comment[state->length++] = len; |
770 |
} while (len && copy < have); |
771 |
if (state->flags & 0x0200) |
772 |
state->check = crc32(state->check, next, copy); |
773 |
have -= copy; |
774 |
next += copy; |
775 |
if (len) goto inf_leave; |
776 |
} |
777 |
else if (state->head != Z_NULL) |
778 |
state->head->comment = Z_NULL; |
779 |
state->mode = HCRC; |
780 |
case HCRC: |
781 |
if (state->flags & 0x0200) { |
782 |
NEEDBITS(16); |
783 |
if (hold != (state->check & 0xffff)) { |
784 |
strm->msg = (char *)"header crc mismatch"; |
785 |
state->mode = BAD; |
786 |
break; |
787 |
} |
788 |
INITBITS(); |
789 |
} |
790 |
if (state->head != Z_NULL) { |
791 |
state->head->hcrc = (int)((state->flags >> 9) & 1); |
792 |
state->head->done = 1; |
793 |
} |
794 |
strm->adler = state->check = crc32(0L, Z_NULL, 0); |
795 |
state->mode = TYPE; |
796 |
break; |
797 |
#endif |
798 |
case DICTID: |
799 |
NEEDBITS(32); |
800 |
strm->adler = state->check = REVERSE(hold); |
801 |
INITBITS(); |
802 |
state->mode = DICT; |
803 |
case DICT: |
804 |
if (state->havedict == 0) { |
805 |
RESTORE(); |
806 |
return Z_NEED_DICT; |
807 |
} |
808 |
strm->adler = state->check = adler32(0L, Z_NULL, 0); |
809 |
state->mode = TYPE; |
810 |
case TYPE: |
811 |
if (flush == Z_BLOCK || flush == Z_TREES) goto inf_leave; |
812 |
case TYPEDO: |
813 |
if (state->last) { |
814 |
BYTEBITS(); |
815 |
state->mode = CHECK; |
816 |
break; |
817 |
} |
818 |
NEEDBITS(3); |
819 |
state->last = BITS(1); |
820 |
DROPBITS(1); |
821 |
switch (BITS(2)) { |
822 |
case 0: /* stored block */ |
823 |
Tracev((stderr, "inflate: stored block%s\n", |
824 |
state->last ? " (last)" : "")); |
825 |
state->mode = STORED; |
826 |
break; |
827 |
case 1: /* fixed block */ |
828 |
fixedtables(state); |
829 |
Tracev((stderr, "inflate: fixed codes block%s\n", |
830 |
state->last ? " (last)" : "")); |
831 |
state->mode = LEN_; /* decode codes */ |
832 |
if (flush == Z_TREES) { |
833 |
DROPBITS(2); |
834 |
goto inf_leave; |
835 |
} |
836 |
break; |
837 |
case 2: /* dynamic block */ |
838 |
Tracev((stderr, "inflate: dynamic codes block%s\n", |
839 |
state->last ? " (last)" : "")); |
840 |
state->mode = TABLE; |
841 |
break; |
842 |
case 3: |
843 |
strm->msg = (char *)"invalid block type"; |
844 |
state->mode = BAD; |
845 |
} |
846 |
DROPBITS(2); |
847 |
break; |
848 |
case STORED: |
849 |
BYTEBITS(); /* go to byte boundary */ |
850 |
NEEDBITS(32); |
851 |
if ((hold & 0xffff) != ((hold >> 16) ^ 0xffff)) { |
852 |
strm->msg = (char *)"invalid stored block lengths"; |
853 |
state->mode = BAD; |
854 |
break; |
855 |
} |
856 |
state->length = (unsigned)hold & 0xffff; |
857 |
Tracev((stderr, "inflate: stored length %u\n", |
858 |
state->length)); |
859 |
INITBITS(); |
860 |
state->mode = COPY_; |
861 |
if (flush == Z_TREES) goto inf_leave; |
862 |
case COPY_: |
863 |
state->mode = COPY; |
864 |
case COPY: |
865 |
copy = state->length; |
866 |
if (copy) { |
867 |
if (copy > have) copy = have; |
868 |
if (copy > left) copy = left; |
869 |
if (copy == 0) goto inf_leave; |
870 |
zmemcpy(put, next, copy); |
871 |
have -= copy; |
872 |
next += copy; |
873 |
left -= copy; |
874 |
put += copy; |
875 |
state->length -= copy; |
876 |
break; |
877 |
} |
878 |
Tracev((stderr, "inflate: stored end\n")); |
879 |
state->mode = TYPE; |
880 |
break; |
881 |
case TABLE: |
882 |
NEEDBITS(14); |
883 |
state->nlen = BITS(5) + 257; |
884 |
DROPBITS(5); |
885 |
state->ndist = BITS(5) + 1; |
886 |
DROPBITS(5); |
887 |
state->ncode = BITS(4) + 4; |
888 |
DROPBITS(4); |
889 |
#ifndef PKZIP_BUG_WORKAROUND |
890 |
if (state->nlen > 286 || state->ndist > 30) { |
891 |
strm->msg = (char *)"too many length or distance symbols"; |
892 |
state->mode = BAD; |
893 |
break; |
894 |
} |
895 |
#endif |
896 |
Tracev((stderr, "inflate: table sizes ok\n")); |
897 |
state->have = 0; |
898 |
state->mode = LENLENS; |
899 |
case LENLENS: |
900 |
while (state->have < state->ncode) { |
901 |
NEEDBITS(3); |
902 |
state->lens[order[state->have++]] = (unsigned short)BITS(3); |
903 |
DROPBITS(3); |
904 |
} |
905 |
while (state->have < 19) |
906 |
state->lens[order[state->have++]] = 0; |
907 |
state->next = state->codes; |
908 |
state->lencode = (code const FAR *)(state->next); |
909 |
state->lenbits = 7; |
910 |
ret = inflate_table(CODES, state->lens, 19, &(state->next), |
911 |
&(state->lenbits), state->work); |
912 |
if (ret) { |
913 |
strm->msg = (char *)"invalid code lengths set"; |
914 |
state->mode = BAD; |
915 |
break; |
916 |
} |
917 |
Tracev((stderr, "inflate: code lengths ok\n")); |
918 |
state->have = 0; |
919 |
state->mode = CODELENS; |
920 |
case CODELENS: |
921 |
while (state->have < state->nlen + state->ndist) { |
922 |
for (;;) { |
923 |
here = state->lencode[BITS(state->lenbits)]; |
924 |
if ((unsigned)(here.bits) <= bits) break; |
925 |
PULLBYTE(); |
926 |
} |
927 |
if (here.val < 16) { |
928 |
NEEDBITS(here.bits); |
929 |
DROPBITS(here.bits); |
930 |
state->lens[state->have++] = here.val; |
931 |
} |
932 |
else { |
933 |
if (here.val == 16) { |
934 |
NEEDBITS(here.bits + 2); |
935 |
DROPBITS(here.bits); |
936 |
if (state->have == 0) { |
937 |
strm->msg = (char *)"invalid bit length repeat"; |
938 |
state->mode = BAD; |
939 |
break; |
940 |
} |
941 |
len = state->lens[state->have - 1]; |
942 |
copy = 3 + BITS(2); |
943 |
DROPBITS(2); |
944 |
} |
945 |
else if (here.val == 17) { |
946 |
NEEDBITS(here.bits + 3); |
947 |
DROPBITS(here.bits); |
948 |
len = 0; |
949 |
copy = 3 + BITS(3); |
950 |
DROPBITS(3); |
951 |
} |
952 |
else { |
953 |
NEEDBITS(here.bits + 7); |
954 |
DROPBITS(here.bits); |
955 |
len = 0; |
956 |
copy = 11 + BITS(7); |
957 |
DROPBITS(7); |
958 |
} |
959 |
if (state->have + copy > state->nlen + state->ndist) { |
960 |
strm->msg = (char *)"invalid bit length repeat"; |
961 |
state->mode = BAD; |
962 |
break; |
963 |
} |
964 |
while (copy--) |
965 |
state->lens[state->have++] = (unsigned short)len; |
966 |
} |
967 |
} |
968 |
|
969 |
/* handle error breaks in while */ |
970 |
if (state->mode == BAD) break; |
971 |
|
972 |
/* check for end-of-block code (better have one) */ |
973 |
if (state->lens[256] == 0) { |
974 |
strm->msg = (char *)"invalid code -- missing end-of-block"; |
975 |
state->mode = BAD; |
976 |
break; |
977 |
} |
978 |
|
979 |
/* build code tables -- note: do not change the lenbits or distbits |
980 |
values here (9 and 6) without reading the comments in inftrees.h |
981 |
concerning the ENOUGH constants, which depend on those values */ |
982 |
state->next = state->codes; |
983 |
state->lencode = (code const FAR *)(state->next); |
984 |
state->lenbits = 9; |
985 |
ret = inflate_table(LENS, state->lens, state->nlen, &(state->next), |
986 |
&(state->lenbits), state->work); |
987 |
if (ret) { |
988 |
strm->msg = (char *)"invalid literal/lengths set"; |
989 |
state->mode = BAD; |
990 |
break; |
991 |
} |
992 |
state->distcode = (code const FAR *)(state->next); |
993 |
state->distbits = 6; |
994 |
ret = inflate_table(DISTS, state->lens + state->nlen, state->ndist, |
995 |
&(state->next), &(state->distbits), state->work); |
996 |
if (ret) { |
997 |
strm->msg = (char *)"invalid distances set"; |
998 |
state->mode = BAD; |
999 |
break; |
1000 |
} |
1001 |
Tracev((stderr, "inflate: codes ok\n")); |
1002 |
state->mode = LEN_; |
1003 |
if (flush == Z_TREES) goto inf_leave; |
1004 |
case LEN_: |
1005 |
state->mode = LEN; |
1006 |
case LEN: |
1007 |
if (have >= 6 && left >= 258) { |
1008 |
RESTORE(); |
1009 |
inflate_fast(strm, out); |
1010 |
LOAD(); |
1011 |
if (state->mode == TYPE) |
1012 |
state->back = -1; |
1013 |
break; |
1014 |
} |
1015 |
state->back = 0; |
1016 |
for (;;) { |
1017 |
here = state->lencode[BITS(state->lenbits)]; |
1018 |
if ((unsigned)(here.bits) <= bits) break; |
1019 |
PULLBYTE(); |
1020 |
} |
1021 |
if (here.op && (here.op & 0xf0) == 0) { |
1022 |
last = here; |
1023 |
for (;;) { |
1024 |
here = state->lencode[last.val + |
1025 |
(BITS(last.bits + last.op) >> last.bits)]; |
1026 |
if ((unsigned)(last.bits + here.bits) <= bits) break; |
1027 |
PULLBYTE(); |
1028 |
} |
1029 |
DROPBITS(last.bits); |
1030 |
state->back += last.bits; |
1031 |
} |
1032 |
DROPBITS(here.bits); |
1033 |
state->back += here.bits; |
1034 |
state->length = (unsigned)here.val; |
1035 |
if ((int)(here.op) == 0) { |
1036 |
Tracevv((stderr, here.val >= 0x20 && here.val < 0x7f ? |
1037 |
"inflate: literal '%c'\n" : |
1038 |
"inflate: literal 0x%02x\n", here.val)); |
1039 |
state->mode = LIT; |
1040 |
break; |
1041 |
} |
1042 |
if (here.op & 32) { |
1043 |
Tracevv((stderr, "inflate: end of block\n")); |
1044 |
state->back = -1; |
1045 |
state->mode = TYPE; |
1046 |
break; |
1047 |
} |
1048 |
if (here.op & 64) { |
1049 |
strm->msg = (char *)"invalid literal/length code"; |
1050 |
state->mode = BAD; |
1051 |
break; |
1052 |
} |
1053 |
state->extra = (unsigned)(here.op) & 15; |
1054 |
state->mode = LENEXT; |
1055 |
case LENEXT: |
1056 |
if (state->extra) { |
1057 |
NEEDBITS(state->extra); |
1058 |
state->length += BITS(state->extra); |
1059 |
DROPBITS(state->extra); |
1060 |
state->back += state->extra; |
1061 |
} |
1062 |
Tracevv((stderr, "inflate: length %u\n", state->length)); |
1063 |
state->was = state->length; |
1064 |
state->mode = DIST; |
1065 |
case DIST: |
1066 |
for (;;) { |
1067 |
here = state->distcode[BITS(state->distbits)]; |
1068 |
if ((unsigned)(here.bits) <= bits) break; |
1069 |
PULLBYTE(); |
1070 |
} |
1071 |
if ((here.op & 0xf0) == 0) { |
1072 |
last = here; |
1073 |
for (;;) { |
1074 |
here = state->distcode[last.val + |
1075 |
(BITS(last.bits + last.op) >> last.bits)]; |
1076 |
if ((unsigned)(last.bits + here.bits) <= bits) break; |
1077 |
PULLBYTE(); |
1078 |
} |
1079 |
DROPBITS(last.bits); |
1080 |
state->back += last.bits; |
1081 |
} |
1082 |
DROPBITS(here.bits); |
1083 |
state->back += here.bits; |
1084 |
if (here.op & 64) { |
1085 |
strm->msg = (char *)"invalid distance code"; |
1086 |
state->mode = BAD; |
1087 |
break; |
1088 |
} |
1089 |
state->offset = (unsigned)here.val; |
1090 |
state->extra = (unsigned)(here.op) & 15; |
1091 |
state->mode = DISTEXT; |
1092 |
case DISTEXT: |
1093 |
if (state->extra) { |
1094 |
NEEDBITS(state->extra); |
1095 |
state->offset += BITS(state->extra); |
1096 |
DROPBITS(state->extra); |
1097 |
state->back += state->extra; |
1098 |
} |
1099 |
#ifdef INFLATE_STRICT |
1100 |
if (state->offset > state->dmax) { |
1101 |
strm->msg = (char *)"invalid distance too far back"; |
1102 |
state->mode = BAD; |
1103 |
break; |
1104 |
} |
1105 |
#endif |
1106 |
Tracevv((stderr, "inflate: distance %u\n", state->offset)); |
1107 |
state->mode = MATCH; |
1108 |
case MATCH: |
1109 |
if (left == 0) goto inf_leave; |
1110 |
copy = out - left; |
1111 |
if (state->offset > copy) { /* copy from window */ |
1112 |
copy = state->offset - copy; |
1113 |
if (copy > state->whave) { |
1114 |
if (state->sane) { |
1115 |
strm->msg = (char *)"invalid distance too far back"; |
1116 |
state->mode = BAD; |
1117 |
break; |
1118 |
} |
1119 |
#ifdef INFLATE_ALLOW_INVALID_DISTANCE_TOOFAR_ARRR |
1120 |
Trace((stderr, "inflate.c too far\n")); |
1121 |
copy -= state->whave; |
1122 |
if (copy > state->length) copy = state->length; |
1123 |
if (copy > left) copy = left; |
1124 |
left -= copy; |
1125 |
state->length -= copy; |
1126 |
do { |
1127 |
*put++ = 0; |
1128 |
} while (--copy); |
1129 |
if (state->length == 0) state->mode = LEN; |
1130 |
break; |
1131 |
#endif |
1132 |
} |
1133 |
if (copy > state->wnext) { |
1134 |
copy -= state->wnext; |
1135 |
from = state->window + (state->wsize - copy); |
1136 |
} |
1137 |
else |
1138 |
from = state->window + (state->wnext - copy); |
1139 |
if (copy > state->length) copy = state->length; |
1140 |
} |
1141 |
else { /* copy from output */ |
1142 |
from = put - state->offset; |
1143 |
copy = state->length; |
1144 |
} |
1145 |
if (copy > left) copy = left; |
1146 |
left -= copy; |
1147 |
state->length -= copy; |
1148 |
do { |
1149 |
*put++ = *from++; |
1150 |
} while (--copy); |
1151 |
if (state->length == 0) state->mode = LEN; |
1152 |
break; |
1153 |
case LIT: |
1154 |
if (left == 0) goto inf_leave; |
1155 |
*put++ = (unsigned char)(state->length); |
1156 |
left--; |
1157 |
state->mode = LEN; |
1158 |
break; |
1159 |
case CHECK: |
1160 |
if (state->wrap) { |
1161 |
NEEDBITS(32); |
1162 |
out -= left; |
1163 |
strm->total_out += out; |
1164 |
state->total += out; |
1165 |
if (out) |
1166 |
strm->adler = state->check = |
1167 |
UPDATE(state->check, put - out, out); |
1168 |
out = left; |
1169 |
if (( |
1170 |
#ifdef GUNZIP |
1171 |
state->flags ? hold : |
1172 |
#endif |
1173 |
REVERSE(hold)) != state->check) { |
1174 |
strm->msg = (char *)"incorrect data check"; |
1175 |
state->mode = BAD; |
1176 |
break; |
1177 |
} |
1178 |
INITBITS(); |
1179 |
Tracev((stderr, "inflate: check matches trailer\n")); |
1180 |
} |
1181 |
#ifdef GUNZIP |
1182 |
state->mode = LENGTH; |
1183 |
case LENGTH: |
1184 |
if (state->wrap && state->flags) { |
1185 |
NEEDBITS(32); |
1186 |
if (hold != (state->total & 0xffffffffUL)) { |
1187 |
strm->msg = (char *)"incorrect length check"; |
1188 |
state->mode = BAD; |
1189 |
break; |
1190 |
} |
1191 |
INITBITS(); |
1192 |
Tracev((stderr, "inflate: length matches trailer\n")); |
1193 |
} |
1194 |
#endif |
1195 |
state->mode = DONE; |
1196 |
case DONE: |
1197 |
ret = Z_STREAM_END; |
1198 |
goto inf_leave; |
1199 |
case BAD: |
1200 |
ret = Z_DATA_ERROR; |
1201 |
goto inf_leave; |
1202 |
case MEM: |
1203 |
return Z_MEM_ERROR; |
1204 |
case SYNC: |
1205 |
default: |
1206 |
return Z_STREAM_ERROR; |
1207 |
} |
1208 |
|
1209 |
/* |
1210 |
Return from inflate(), updating the total counts and the check value. |
1211 |
If there was no progress during the inflate() call, return a buffer |
1212 |
error. Call updatewindow() to create and/or update the window state. |
1213 |
Note: a memory error from inflate() is non-recoverable. |
1214 |
*/ |
1215 |
inf_leave: |
1216 |
RESTORE(); |
1217 |
if (state->wsize || (state->mode < CHECK && out != strm->avail_out)) |
1218 |
if (updatewindow(strm, out)) { |
1219 |
state->mode = MEM; |
1220 |
return Z_MEM_ERROR; |
1221 |
} |
1222 |
in -= strm->avail_in; |
1223 |
out -= strm->avail_out; |
1224 |
strm->total_in += in; |
1225 |
strm->total_out += out; |
1226 |
state->total += out; |
1227 |
if (state->wrap && out) |
1228 |
strm->adler = state->check = |
1229 |
UPDATE(state->check, strm->next_out - out, out); |
1230 |
strm->data_type = state->bits + (state->last ? 64 : 0) + |
1231 |
(state->mode == TYPE ? 128 : 0) + |
1232 |
(state->mode == LEN_ || state->mode == COPY_ ? 256 : 0); |
1233 |
if (((in == 0 && out == 0) || flush == Z_FINISH) && ret == Z_OK) |
1234 |
ret = Z_BUF_ERROR; |
1235 |
return ret; |
1236 |
} |
1237 |
|
1238 |
int ZEXPORT inflateEnd(strm) |
1239 |
z_streamp strm; |
1240 |
{ |
1241 |
struct inflate_state FAR *state; |
1242 |
if (strm == Z_NULL || strm->state == Z_NULL || strm->zfree == (free_func)0) |
1243 |
return Z_STREAM_ERROR; |
1244 |
state = (struct inflate_state FAR *)strm->state; |
1245 |
if (state->window != Z_NULL) ZFREE(strm, state->window); |
1246 |
ZFREE(strm, strm->state); |
1247 |
strm->state = Z_NULL; |
1248 |
Tracev((stderr, "inflate: end\n")); |
1249 |
return Z_OK; |
1250 |
} |
1251 |
|
1252 |
int ZEXPORT inflateSetDictionary(strm, dictionary, dictLength) |
1253 |
z_streamp strm; |
1254 |
const Bytef *dictionary; |
1255 |
uInt dictLength; |
1256 |
{ |
1257 |
struct inflate_state FAR *state; |
1258 |
unsigned long id; |
1259 |
|
1260 |
/* check state */ |
1261 |
if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; |
1262 |
state = (struct inflate_state FAR *)strm->state; |
1263 |
if (state->wrap != 0 && state->mode != DICT) |
1264 |
return Z_STREAM_ERROR; |
1265 |
|
1266 |
/* check for correct dictionary id */ |
1267 |
if (state->mode == DICT) { |
1268 |
id = adler32(0L, Z_NULL, 0); |
1269 |
id = adler32(id, dictionary, dictLength); |
1270 |
if (id != state->check) |
1271 |
return Z_DATA_ERROR; |
1272 |
} |
1273 |
|
1274 |
/* copy dictionary to window */ |
1275 |
if (updatewindow(strm, strm->avail_out)) { |
1276 |
state->mode = MEM; |
1277 |
return Z_MEM_ERROR; |
1278 |
} |
1279 |
if (dictLength > state->wsize) { |
1280 |
zmemcpy(state->window, dictionary + dictLength - state->wsize, |
1281 |
state->wsize); |
1282 |
state->whave = state->wsize; |
1283 |
} |
1284 |
else { |
1285 |
zmemcpy(state->window + state->wsize - dictLength, dictionary, |
1286 |
dictLength); |
1287 |
state->whave = dictLength; |
1288 |
} |
1289 |
state->havedict = 1; |
1290 |
Tracev((stderr, "inflate: dictionary set\n")); |
1291 |
return Z_OK; |
1292 |
} |
1293 |
|
1294 |
int ZEXPORT inflateGetHeader(strm, head) |
1295 |
z_streamp strm; |
1296 |
gz_headerp head; |
1297 |
{ |
1298 |
struct inflate_state FAR *state; |
1299 |
|
1300 |
/* check state */ |
1301 |
if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; |
1302 |
state = (struct inflate_state FAR *)strm->state; |
1303 |
if ((state->wrap & 2) == 0) return Z_STREAM_ERROR; |
1304 |
|
1305 |
/* save header structure */ |
1306 |
state->head = head; |
1307 |
head->done = 0; |
1308 |
return Z_OK; |
1309 |
} |
1310 |
|
1311 |
/* |
1312 |
Search buf[0..len-1] for the pattern: 0, 0, 0xff, 0xff. Return when found |
1313 |
or when out of input. When called, *have is the number of pattern bytes |
1314 |
found in order so far, in 0..3. On return *have is updated to the new |
1315 |
state. If on return *have equals four, then the pattern was found and the |
1316 |
return value is how many bytes were read including the last byte of the |
1317 |
pattern. If *have is less than four, then the pattern has not been found |
1318 |
yet and the return value is len. In the latter case, syncsearch() can be |
1319 |
called again with more data and the *have state. *have is initialized to |
1320 |
zero for the first call. |
1321 |
*/ |
1322 |
local unsigned syncsearch(have, buf, len) |
1323 |
unsigned FAR *have; |
1324 |
unsigned char FAR *buf; |
1325 |
unsigned len; |
1326 |
{ |
1327 |
unsigned got; |
1328 |
unsigned next; |
1329 |
|
1330 |
got = *have; |
1331 |
next = 0; |
1332 |
while (next < len && got < 4) { |
1333 |
if ((int)(buf[next]) == (got < 2 ? 0 : 0xff)) |
1334 |
got++; |
1335 |
else if (buf[next]) |
1336 |
got = 0; |
1337 |
else |
1338 |
got = 4 - got; |
1339 |
next++; |
1340 |
} |
1341 |
*have = got; |
1342 |
return next; |
1343 |
} |
1344 |
|
1345 |
int ZEXPORT inflateSync(strm) |
1346 |
z_streamp strm; |
1347 |
{ |
1348 |
unsigned len; /* number of bytes to look at or looked at */ |
1349 |
unsigned long in, out; /* temporary to save total_in and total_out */ |
1350 |
unsigned char buf[4]; /* to restore bit buffer to byte string */ |
1351 |
struct inflate_state FAR *state; |
1352 |
|
1353 |
/* check parameters */ |
1354 |
if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; |
1355 |
state = (struct inflate_state FAR *)strm->state; |
1356 |
if (strm->avail_in == 0 && state->bits < 8) return Z_BUF_ERROR; |
1357 |
|
1358 |
/* if first time, start search in bit buffer */ |
1359 |
if (state->mode != SYNC) { |
1360 |
state->mode = SYNC; |
1361 |
state->hold <<= state->bits & 7; |
1362 |
state->bits -= state->bits & 7; |
1363 |
len = 0; |
1364 |
while (state->bits >= 8) { |
1365 |
buf[len++] = (unsigned char)(state->hold); |
1366 |
state->hold >>= 8; |
1367 |
state->bits -= 8; |
1368 |
} |
1369 |
state->have = 0; |
1370 |
syncsearch(&(state->have), buf, len); |
1371 |
} |
1372 |
|
1373 |
/* search available input */ |
1374 |
len = syncsearch(&(state->have), strm->next_in, strm->avail_in); |
1375 |
strm->avail_in -= len; |
1376 |
strm->next_in += len; |
1377 |
strm->total_in += len; |
1378 |
|
1379 |
/* return no joy or set up to restart inflate() on a new block */ |
1380 |
if (state->have != 4) return Z_DATA_ERROR; |
1381 |
in = strm->total_in; out = strm->total_out; |
1382 |
inflateReset(strm); |
1383 |
strm->total_in = in; strm->total_out = out; |
1384 |
state->mode = TYPE; |
1385 |
return Z_OK; |
1386 |
} |
1387 |
|
1388 |
/* |
1389 |
Returns true if inflate is currently at the end of a block generated by |
1390 |
Z_SYNC_FLUSH or Z_FULL_FLUSH. This function is used by one PPP |
1391 |
implementation to provide an additional safety check. PPP uses |
1392 |
Z_SYNC_FLUSH but removes the length bytes of the resulting empty stored |
1393 |
block. When decompressing, PPP checks that at the end of input packet, |
1394 |
inflate is waiting for these length bytes. |
1395 |
*/ |
1396 |
int ZEXPORT inflateSyncPoint(strm) |
1397 |
z_streamp strm; |
1398 |
{ |
1399 |
struct inflate_state FAR *state; |
1400 |
|
1401 |
if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; |
1402 |
state = (struct inflate_state FAR *)strm->state; |
1403 |
return state->mode == STORED && state->bits == 0; |
1404 |
} |
1405 |
|
1406 |
int ZEXPORT inflateCopy(dest, source) |
1407 |
z_streamp dest; |
1408 |
z_streamp source; |
1409 |
{ |
1410 |
struct inflate_state FAR *state; |
1411 |
struct inflate_state FAR *copy; |
1412 |
unsigned char FAR *window; |
1413 |
unsigned wsize; |
1414 |
|
1415 |
/* check input */ |
1416 |
if (dest == Z_NULL || source == Z_NULL || source->state == Z_NULL || |
1417 |
source->zalloc == (alloc_func)0 || source->zfree == (free_func)0) |
1418 |
return Z_STREAM_ERROR; |
1419 |
state = (struct inflate_state FAR *)source->state; |
1420 |
|
1421 |
/* allocate space */ |
1422 |
copy = (struct inflate_state FAR *) |
1423 |
ZALLOC(source, 1, sizeof(struct inflate_state)); |
1424 |
if (copy == Z_NULL) return Z_MEM_ERROR; |
1425 |
window = Z_NULL; |
1426 |
if (state->window != Z_NULL) { |
1427 |
window = (unsigned char FAR *) |
1428 |
ZALLOC(source, 1U << state->wbits, sizeof(unsigned char)); |
1429 |
if (window == Z_NULL) { |
1430 |
ZFREE(source, copy); |
1431 |
return Z_MEM_ERROR; |
1432 |
} |
1433 |
} |
1434 |
|
1435 |
/* copy state */ |
1436 |
zmemcpy(dest, source, sizeof(z_stream)); |
1437 |
zmemcpy(copy, state, sizeof(struct inflate_state)); |
1438 |
if (state->lencode >= state->codes && |
1439 |
state->lencode <= state->codes + ENOUGH - 1) { |
1440 |
copy->lencode = copy->codes + (state->lencode - state->codes); |
1441 |
copy->distcode = copy->codes + (state->distcode - state->codes); |
1442 |
} |
1443 |
copy->next = copy->codes + (state->next - state->codes); |
1444 |
if (window != Z_NULL) { |
1445 |
wsize = 1U << state->wbits; |
1446 |
zmemcpy(window, state->window, wsize); |
1447 |
} |
1448 |
copy->window = window; |
1449 |
dest->state = (struct internal_state FAR *)copy; |
1450 |
return Z_OK; |
1451 |
} |
1452 |
|
1453 |
int ZEXPORT inflateUndermine(strm, subvert) |
1454 |
z_streamp strm; |
1455 |
int subvert; |
1456 |
{ |
1457 |
struct inflate_state FAR *state; |
1458 |
|
1459 |
if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; |
1460 |
state = (struct inflate_state FAR *)strm->state; |
1461 |
state->sane = !subvert; |
1462 |
#ifdef INFLATE_ALLOW_INVALID_DISTANCE_TOOFAR_ARRR |
1463 |
return Z_OK; |
1464 |
#else |
1465 |
state->sane = 1; |
1466 |
return Z_DATA_ERROR; |
1467 |
#endif |
1468 |
} |
1469 |
|
1470 |
long ZEXPORT inflateMark(strm) |
1471 |
z_streamp strm; |
1472 |
{ |
1473 |
struct inflate_state FAR *state; |
1474 |
|
1475 |
if (strm == Z_NULL || strm->state == Z_NULL) return -1L << 16; |
1476 |
state = (struct inflate_state FAR *)strm->state; |
1477 |
return ((long)(state->back) << 16) + |
1478 |
(state->mode == COPY ? state->length : |
1479 |
(state->mode == MATCH ? state->was - state->length : 0)); |
1480 |
} |