1 /*
2 Copyright (C) 2000 Masanao Izumo <mo@goice.co.jp>
3
4 This program is free software; you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation; either version 2 of the License, or
7 (at your option) any later version.
8
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
13
14 You should have received a copy of the GNU General Public License
15 along with this program; if not, write to the Free Software
16 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
17 */
18
19 /* inflate.c -- Not copyrighted 1992 by Mark Adler
20 version c10p1, 10 January 1993 */
21
22 /* You can do whatever you like with this source file, though I would
23 prefer that if you modify it and redistribute it that you include
24 comments to that effect with your name and the date. Thank you.
25 [The history has been moved to the file ChangeLog.]
26 */
27
28 /*
29 Inflate deflated (PKZIP's method 8 compressed) data. The compression
30 method searches for as much of the current string of bytes (up to a
31 length of 258) in the previous 32K bytes. If it doesn't find any
32 matches (of at least length 3), it codes the next byte. Otherwise, it
33 codes the length of the matched string and its distance backwards from
34 the current position. There is a single Huffman code that codes both
35 single bytes (called "literals") and match lengths. A second Huffman
36 code codes the distance information, which follows a length code. Each
37 length or distance code actually represents a base value and a number
38 of "extra" (sometimes zero) bits to get to add to the base value. At
39 the end of each deflated block is a special end-of-block (EOB) literal/
40 length code. The decoding process is basically: get a literal/length
41 code; if EOB then done; if a literal, emit the decoded byte; if a
42 length then get the distance and emit the referred-to bytes from the
43 sliding window of previously emitted data.
44
45 There are (currently) three kinds of inflate blocks: stored, fixed, and
46 dynamic. The compressor outputs a chunk of data at a time and decides
47 which method to use on a chunk-by-chunk basis. A chunk might typically
48 be 32K to 64K, uncompressed. If the chunk is uncompressible, then the
49 "stored" method is used. In this case, the bytes are simply stored as
50 is, eight bits per byte, with none of the above coding. The bytes are
51 preceded by a count, since there is no longer an EOB code.
52
53 If the data are compressible, then either the fixed or dynamic methods
54 are used. In the dynamic method, the compressed data are preceded by
55 an encoding of the literal/length and distance Huffman codes that are
56 to be used to decode this block. The representation is itself Huffman
57 coded, and so is preceded by a description of that code. These code
58 descriptions take up a little space, and so for small blocks, there is
59 a predefined set of codes, called the fixed codes. The fixed method is
60 used if the block ends up smaller that way (usually for quite small
61 chunks); otherwise the dynamic method is used. In the latter case, the
62 codes are customized to the probabilities in the current block and so
63 can code it much better than the pre-determined fixed codes can.
64
65 The Huffman codes themselves are decoded using a multi-level table
66 lookup, in order to maximize the speed of decoding plus the speed of
67 building the decoding tables. See the comments below that precede the
68 lbits and dbits tuning parameters.
69 */
70
71
72 /*
73 Notes beyond the 1.93a appnote.txt:
74
75 1. Distance pointers never point before the beginning of the output
76 stream.
77 2. Distance pointers can point back across blocks, up to 32k away.
78 3. There is an implied maximum of 7 bits for the bit length table and
79 15 bits for the actual data.
80 4. If only one code exists, then it is encoded using one bit. (Zero
81 would be more efficient, but perhaps a little confusing.) If two
82 codes exist, they are coded using one bit each (0 and 1).
83 5. There is no way of sending zero distance codes--a dummy must be
84 sent if there are none. (History: a pre 2.0 version of PKZIP would
85 store blocks with no distance codes, but this was discovered to be
86 too harsh a criterion.) Valid only for 1.93a. 2.04c does allow
87 zero distance codes, which is sent as one code of zero bits in
88 length.
89 6. There are up to 286 literal/length codes. Code 256 represents the
90 end-of-block. Note however that the static length tree defines
91 288 codes just to fill out the Huffman codes. Codes 286 and 287
92 cannot be used though, since there is no length base or extra bits
93 defined for them. Similarily, there are up to 30 distance codes.
94 However, static trees define 32 codes (all 5 bits) to fill out the
95 Huffman codes, but the last two had better not show up in the data.
96 7. Unzip can check dynamic Huffman blocks for complete code sets.
97 The exception is that a single code would not be complete (see #4).
98 8. The five bits following the block type is really the number of
99 literal codes sent minus 257.
100 9. Length codes 8,16,16 are interpreted as 13 length codes of 8 bits
101 (1+6+6). Therefore, to output three times the length, you output
102 three codes (1+1+1), whereas to output four times the same length,
103 you only need two codes (1+3). Hmm.
104 10. In the tree reconstruction algorithm, Code = Code + Increment
105 only if BitLength(i) is not zero. (Pretty obvious.)
106 11. Correction: 4 Bits: # of Bit Length codes - 4 (4 - 19)
107 12. Note: length code 284 can represent 227-258, but length code 285
108 really is 258. The last length deserves its own, short code
109 since it gets used a lot in very redundant files. The length
110 258 is special since 258 - 3 (the min match length) is 255.
111 13. The literal/length and distance code bit lengths are read as a
112 single stream of lengths. It is possible (and advantageous) for
113 a repeat code (16, 17, or 18) to go across the boundary between
114 the two sets of lengths.
115 */
116
117 #include "config.h"
118 #include <stdio.h>
119 #include <stdlib.h>
120 #ifndef NO_STRING_H
121 #include <string.h>
122 #else
123 #include <strings.h>
124 #endif
125 #include "libarc/mblock.h"
126 #include "zip.h"
127 #define local static
128
129 /* Save to local */
130 #define BITS_SAVE \
131 ulg bit_buf = decoder->bit_buf; \
132 ulg bit_len = decoder->bit_len;
133
134 /* Restore to decoder */
135 #define BITS_RESTORE \
136 decoder->bit_buf = bit_buf; \
137 decoder->bit_len = bit_len;
138
139 #define MASK_BITS(n) ((((ulg)1)<<(n))-1)
140 #define GET_BYTE() (decoder->inptr < decoder->insize ? decoder->inbuf[decoder->inptr++] : fill_inbuf(decoder))
141 #define NEEDBITS(n) {while(bit_len<(n)){bit_buf|=((ulg)GET_BYTE())<<bit_len;bit_len+=8;}}
142 #define GETBITS(n) (bit_buf & MASK_BITS(n))
143 #define DUMPBITS(n) {bit_buf>>=(n);bit_len-=(n);}
144
145 /* variables */
146 struct _InflateHandler
147 {
148 void *user_val;
149 long (* read_func)(char *buf, long size, void *user_val);
150
151 uch slide[2L * WSIZE];
152 uch inbuf[INBUFSIZ + INBUF_EXTRA];
153 unsigned wp; /* current position in slide */
154 unsigned insize; /* valid bytes in inbuf */
155 unsigned inptr; /* index of next byte to be processed in inbuf */
156 struct huft *fixed_tl; /* inflate static */
157 struct huft *fixed_td; /* inflate static */
158 int fixed_bl, fixed_bd; /* inflate static */
159 ulg bit_buf; /* bit buffer */
160 ulg bit_len; /* bits in bit buffer */
161 int method;
162 int eof;
163 unsigned copy_leng;
164 unsigned copy_dist;
165 struct huft *tl, *td; /* literal/length and distance decoder tables */
166 int bl, bd; /* number of bits decoded by tl[] and td[] */
167 MBlockList pool; /* memory buffer for tl, td */
168 };
169
170 /* Function prototypes */
171 local int fill_inbuf(InflateHandler);
172 local int huft_free(struct huft *);
173 local long inflate_codes(InflateHandler, char *, long);
174 local long inflate_stored(InflateHandler, char *, long);
175 local long inflate_fixed(InflateHandler, char *, long);
176 local long inflate_dynamic(InflateHandler, char *, long);
177 local void inflate_start(InflateHandler);
178
179 /* The inflate algorithm uses a sliding 32K byte window on the uncompressed
180 stream to find repeated byte strings. This is implemented here as a
181 circular buffer. The index is updated simply by incrementing and then
182 and'ing with 0x7fff (32K-1). */
183 /* It is left to other modules to supply the 32K area. It is assumed
184 to be usable as if it were declared "uch slide[32768];" or as just
185 "uch *slide;" and then malloc'ed in the latter case. The definition
186 must be in unzip.h, included above. */
187
188 #define lbits 9 /* bits in base literal/length lookup table */
189 #define dbits 6 /* bits in base distance lookup table */
190
191 /* Tables for deflate from PKZIP's appnote.txt. */
192 local ush cplens[] = { /* Copy lengths for literal codes 257..285 */
193 3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31,
194 35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, 227, 258, 0, 0};
195 /* note: see note #13 above about the 258 in this list. */
196 local ush cplext[] = { /* Extra bits for literal codes 257..285 */
197 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
198 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0, 99, 99}; /* 99==invalid */
199 local ush cpdist[] = { /* Copy offsets for distance codes 0..29 */
200 1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193,
201 257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, 6145,
202 8193, 12289, 16385, 24577};
203 local ush cpdext[] = { /* Extra bits for distance codes */
204 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
205 7, 7, 8, 8, 9, 9, 10, 10, 11, 11,
206 12, 12, 13, 13};
207
208 /*
209 Huffman code decoding is performed using a multi-level table lookup.
210 The fastest way to decode is to simply build a lookup table whose
211 size is determined by the longest code. However, the time it takes
212 to build this table can also be a factor if the data being decoded
213 are not very long. The most common codes are necessarily the
214 shortest codes, so those codes dominate the decoding time, and hence
215 the speed. The idea is you can have a shorter table that decodes the
216 shorter, more probable codes, and then point to subsidiary tables for
217 the longer codes. The time it costs to decode the longer codes is
218 then traded against the time it takes to make longer tables.
219
220 This results of this trade are in the variables lbits and dbits
221 below. lbits is the number of bits the first level table for literal/
222 length codes can decode in one step, and dbits is the same thing for
223 the distance codes. Subsequent tables are also less than or equal to
224 those sizes. These values may be adjusted either when all of the
225 codes are shorter than that, in which case the longest code length in
226 bits is used, or when the shortest code is *longer* than the requested
227 table size, in which case the length of the shortest code in bits is
228 used.
229
230 There are two different values for the two tables, since they code a
231 different number of possibilities each. The literal/length table
232 codes 286 possible values, or in a flat code, a little over eight
233 bits. The distance table codes 30 possible values, or a little less
234 than five bits, flat. The optimum values for speed end up being
235 about one bit more than those, so lbits is 8+1 and dbits is 5+1.
236 The optimum values may differ though from machine to machine, and
237 possibly even between compilers. Your mileage may vary.
238 */
239
240 /* If BMAX needs to be larger than 16, then h and x[] should be ulg. */
241 #define BMAX 16 /* maximum bit length of any code (16 for explode) */
242 #define N_MAX 288 /* maximum number of codes in any set */
243
huft_build(unsigned * b,unsigned n,unsigned s,ush * d,ush * e,struct huft ** t,int * m,MBlockList * pool)244 int huft_build(
245 unsigned *b, /* code lengths in bits (all assumed <= BMAX) */
246 unsigned n, /* number of codes (assumed <= N_MAX) */
247 unsigned s, /* number of simple-valued codes (0..s-1) */
248 ush *d, /* list of base values for non-simple codes */
249 ush *e, /* list of extra bits for non-simple codes */
250 struct huft **t, /* result: starting table */
251 int *m, /* maximum lookup bits, returns actual */
252 MBlockList *pool) /* memory pool */
253 /* Given a list of code lengths and a maximum table size, make a set of
254 tables to decode that set of codes. Return zero on success, one if
255 the given code set is incomplete (the tables are still built in this
256 case), two if the input is invalid (all zero length codes or an
257 oversubscribed set of lengths), and three if not enough memory.
258 The code with value 256 is special, and the tables are constructed
259 so that no bits beyond that code are fetched when that code is
260 decoded. */
261 {
262 unsigned a; /* counter for codes of length k */
263 unsigned c[BMAX+1]; /* bit length count table */
264 unsigned el; /* length of EOB code (value 256) */
265 unsigned f; /* i repeats in table every f entries */
266 int g; /* maximum code length */
267 int h; /* table level */
268 register unsigned i; /* counter, current code */
269 register unsigned j; /* counter */
270 register int k; /* number of bits in current code */
271 int lx[BMAX+1]; /* memory for l[-1..BMAX-1] */
272 int *l = lx+1; /* stack of bits per table */
273 register unsigned *p; /* pointer into c[], b[], or v[] */
274 register struct huft *q; /* points to current table */
275 struct huft r; /* table entry for structure assignment */
276 struct huft *u[BMAX]; /* table stack */
277 unsigned v[N_MAX]; /* values in order of bit length */
278 register int w; /* bits before this table == (l * h) */
279 unsigned x[BMAX+1]; /* bit offsets, then code stack */
280 unsigned *xp; /* pointer into x */
281 int y; /* number of dummy codes added */
282 unsigned z; /* number of entries in current table */
283
284 /* Generate counts for each bit length */
285 el = n > 256 ? b[256] : BMAX; /* set length of EOB code, if any */
286 memset(c, 0, sizeof(c));
287 p = b;
288 i = n;
289 do
290 {
291 Tracecv(*p, (stderr, (n-i >= ' ' && n-i <= '~' ? "%c %d\n" :
292 "0x%x %d\n"), n-i, *p));
293 c[*p]++; /* assume all entries <= BMAX */
294 p++; /* Can't combine with above line (Solaris bug) */
295 } while(--i);
296 if(c[0] == n) /* null input--all zero length codes */
297 {
298 *t = (struct huft *)NULL;
299 *m = 0;
300 return 0;
301 }
302
303 /* Find minimum and maximum length, bound *m by those */
304 for(j = 1; j <= BMAX; j++)
305 if(c[j])
306 break;
307 k = j; /* minimum code length */
308 if((unsigned)*m < j)
309 *m = j;
310 for(i = BMAX; i; i--)
311 if(c[i])
312 break;
313 g = i; /* maximum code length */
314 if((unsigned)*m > i)
315 *m = i;
316
317 /* Adjust last length count to fill out codes, if needed */
318 for(y = 1 << j; j < i; j++, y <<= 1)
319 if((y -= c[j]) < 0)
320 return 2; /* bad input: more codes than bits */
321 if((y -= c[i]) < 0)
322 return 2;
323 c[i] += y;
324
325 /* Generate starting offsets into the value table for each length */
326 x[1] = j = 0;
327 p = c + 1; xp = x + 2;
328 while(--i) /* note that i == g from above */
329 *xp++ = (j += *p++);
330
331 /* Make a table of values in order of bit lengths */
332 memset(v, 0, sizeof(v));
333 p = b;
334 i = 0;
335 do
336 {
337 if((j = *p++) != 0)
338 v[x[j]++] = i;
339 } while(++i < n);
340 n = x[g]; /* set n to length of v */
341
342 /* Generate the Huffman codes and for each, make the table entries */
343 x[0] = i = 0; /* first Huffman code is zero */
344 p = v; /* grab values in bit order */
345 h = -1; /* no tables yet--level -1 */
346 w = l[-1] = 0; /* no bits decoded yet */
347 u[0] = (struct huft *)NULL; /* just to keep compilers happy */
348 q = (struct huft *)NULL; /* ditto */
349 z = 0; /* ditto */
350
351 /* go through the bit lengths (k already is bits in shortest code) */
352 for(; k <= g; k++)
353 {
354 a = c[k];
355 while(a--)
356 {
357 /* here i is the Huffman code of length k bits for value *p */
358 /* make tables up to required level */
359 while(k > w + l[h])
360 {
361 w += l[h++]; /* add bits already decoded */
362
363 /* compute minimum size table less than or equal to *m bits */
364 z = (z = g - w) > (unsigned)*m ? *m : z; /* upper limit */
365 if((f = 1 << (j = k - w)) > a + 1) /* try a k-w bit table */
366 { /* too few codes for k-w bit table */
367 f -= a + 1; /* deduct codes from patterns left */
368 xp = c + k;
369 while(++j < z)/* try smaller tables up to z bits */
370 {
371 if((f <<= 1) <= *++xp)
372 break; /* enough codes to use up j bits */
373 f -= *xp; /* else deduct codes from patterns */
374 }
375 }
376 if((unsigned)w + j > el && (unsigned)w < el)
377 j = el - w; /* make EOB code end at table */
378 z = 1 << j; /* table entries for j-bit table */
379 l[h] = j; /* set table size in stack */
380
381 /* allocate and link in new table */
382 if(pool == NULL)
383 q = (struct huft *)malloc((z + 1)*sizeof(struct huft));
384 else
385 q = (struct huft *)
386 new_segment(pool, (z + 1)*sizeof(struct huft));
387 if(q == NULL)
388 {
389 if(h && pool == NULL)
390 huft_free(u[0]);
391 return 3; /* not enough memory */
392 }
393
394 *t = q + 1; /* link to list for huft_free() */
395 *(t = &(q->v.t)) = (struct huft *)NULL;
396 u[h] = ++q; /* table starts after link */
397
398 /* connect to last table, if there is one */
399 if(h)
400 {
401 x[h] = i; /* save pattern for backing up */
402 r.b = (uch)l[h-1]; /* bits to dump before this table */
403 r.e = (uch)(16 + j);/* bits in this table */
404 r.v.t = q; /* pointer to this table */
405 j = (i & ((1 << w) - 1)) >> (w - l[h-1]);
406 u[h-1][j] = r; /* connect to last table */
407 }
408 }
409
410 /* set up table entry in r */
411 r.b = (uch)(k - w);
412 if(p >= v + n)
413 r.e = 99; /* out of values--invalid code */
414 else if(*p < s)
415 {
416 r.e = (uch)(*p < 256 ? 16 : 15); /* 256 is end-of-block code */
417 r.v.n = (ush)*p++; /* simple code is just the value */
418 }
419 else
420 {
421 r.e = (uch)e[*p - s]; /* non-simple--look up in lists */
422 r.v.n = d[*p++ - s];
423 }
424
425 /* fill code-like entries with r */
426 f = 1 << (k - w);
427 for(j = i >> w; j < z; j += f)
428 q[j] = r;
429
430 /* backwards increment the k-bit code i */
431 for(j = 1 << (k - 1); i & j; j >>= 1)
432 i ^= j;
433 i ^= j;
434
435 /* backup over finished tables */
436 while((i & ((1 << w) - 1)) != x[h])
437 w -= l[--h]; /* don't need to update q */
438 }
439 }
440
441 /* return actual size of base table */
442 *m = l[0];
443
444 /* Return true (1) if we were given an incomplete table */
445 return y != 0 && g != 1;
446 }
447
huft_free(struct huft * t)448 local int huft_free(struct huft *t)
449 /* Free the malloc'ed tables built by huft_build(), which makes a linked
450 list of the tables it made, with the links in a dummy first entry of
451 each table. */
452 {
453 register struct huft *p, *q;
454
455 /* Go through linked list, freeing from the malloced (t[-1]) address. */
456 p = t;
457 while(p != (struct huft *)NULL)
458 {
459 q = (--p)->v.t;
460 free((char*)p);
461 p = q;
462 }
463 return 0;
464 }
465
inflate_codes(InflateHandler decoder,char * buff,long size)466 local long inflate_codes(InflateHandler decoder, char *buff, long size)
467 /* inflate (decompress) the codes in a deflated (compressed) block.
468 Return an error code or zero if it all goes ok. */
469 {
470 register unsigned e;/* table entry flag/number of extra bits */
471 struct huft *t; /* pointer to table entry */
472 int n;
473 struct huft *tl, *td;/* literal/length and distance decoder tables */
474 int bl, bd; /* number of bits decoded by tl[] and td[] */
475 unsigned l, w, d;
476 uch *slide;
477
478 BITS_SAVE;
479
480 if(size == 0)
481 return 0;
482
483 slide = decoder->slide;
484 tl = decoder->tl;
485 td = decoder->td;
486 bl = decoder->bl;
487 bd = decoder->bd;
488
489 #ifdef DEBUG
490 if(decoder->copy_leng != 0)
491 {
492 fprintf(stderr, "What ? (decoder->copy_leng = %d)\n",
493 decoder->copy_leng);
494 abort();
495 }
496 #endif /* DEBUG */
497 w = decoder->wp;
498
499 /* inflate the coded data */
500 n = 0;
501 for(;;) /* do until end of block */
502 {
503 NEEDBITS((unsigned)bl);
504 t = tl + GETBITS(bl);
505 e = t->e;
506 while(e > 16)
507 {
508 if(e == 99)
509 return -1;
510 DUMPBITS(t->b);
511 e -= 16;
512 NEEDBITS(e);
513 t = t->v.t + GETBITS(e);
514 e = t->e;
515 }
516 DUMPBITS(t->b);
517
518 if(e == 16) /* then it's a literal */
519 {
520 w &= WSIZE - 1;
521 buff[n++] = slide[w++] = (uch)t->v.n;
522 if(n == size)
523 {
524 decoder->wp = w;
525 BITS_RESTORE;
526 return size;
527 }
528 continue;
529 }
530
531 /* exit if end of block */
532 if(e == 15)
533 break;
534
535 /* it's an EOB or a length */
536
537 /* get length of block to copy */
538 NEEDBITS(e);
539 l = t->v.n + GETBITS(e);
540 DUMPBITS(e);
541
542 /* decode distance of block to copy */
543 NEEDBITS((unsigned)bd);
544 t = td + GETBITS(bd);
545 e = t->e;
546 while(e > 16)
547 {
548 if(e == 99)
549 return -1;
550 DUMPBITS(t->b);
551 e -= 16;
552 NEEDBITS(e);
553 t = t->v.t + GETBITS(e);
554 e = t->e;
555 }
556 DUMPBITS(t->b);
557 NEEDBITS(e);
558 d = w - t->v.n - GETBITS(e);
559 DUMPBITS(e);
560
561 /* do the copy */
562 while(l > 0 && n < size)
563 {
564 l--;
565 d &= WSIZE - 1;
566 w &= WSIZE - 1;
567 buff[n++] = slide[w++] = slide[d++];
568 }
569
570 if(n == size)
571 {
572 decoder->copy_leng = l;
573 decoder->wp = w;
574 decoder->copy_dist = d;
575 BITS_RESTORE;
576 return n;
577 }
578 }
579
580 decoder->wp = w;
581 decoder->method = -1; /* done */
582 BITS_RESTORE;
583 return n;
584 }
585
inflate_stored(InflateHandler decoder,char * buff,long size)586 local long inflate_stored(InflateHandler decoder, char *buff, long size)
587 /* "decompress" an inflated type 0 (stored) block. */
588 {
589 unsigned n, l, w;
590 BITS_SAVE;
591
592 /* go to byte boundary */
593 n = bit_len & 7;
594 DUMPBITS(n);
595
596 /* get the length and its complement */
597 NEEDBITS(16);
598 n = GETBITS(16);
599 DUMPBITS(16);
600 NEEDBITS(16);
601 if(n != (unsigned)((~bit_buf) & 0xffff))
602 {
603 BITS_RESTORE;
604 return -1; /* error in compressed data */
605 }
606 DUMPBITS(16);
607
608 /* read and output the compressed data */
609 decoder->copy_leng = n;
610
611 n = 0;
612 l = decoder->copy_leng;
613 w = decoder->wp;
614 while(l > 0 && n < size)
615 {
616 l--;
617 w &= WSIZE - 1;
618 NEEDBITS(8);
619 buff[n++] = decoder->slide[w++] = (uch)GETBITS(8);
620 DUMPBITS(8);
621 }
622 if(l == 0)
623 decoder->method = -1; /* done */
624 decoder->copy_leng = l;
625 decoder->wp = w;
626 BITS_RESTORE;
627 return (long)n;
628 }
629
inflate_fixed(InflateHandler decoder,char * buff,long size)630 local long inflate_fixed(InflateHandler decoder, char *buff, long size)
631 /* decompress an inflated type 1 (fixed Huffman codes) block. We should
632 either replace this with a custom decoder, or at least precompute the
633 Huffman tables. */
634 {
635 /* if first time, set up tables for fixed blocks */
636 if(decoder->fixed_tl == NULL)
637 {
638 int i; /* temporary variable */
639 unsigned l[288]; /* length list for huft_build */
640
641 /* literal table */
642 for(i = 0; i < 144; i++)
643 l[i] = 8;
644 for(; i < 256; i++)
645 l[i] = 9;
646 for(; i < 280; i++)
647 l[i] = 7;
648 for(; i < 288; i++) /* make a complete, but wrong code set */
649 l[i] = 8;
650 decoder->fixed_bl = 7;
651 if((i = huft_build(l, 288, 257, cplens, cplext,
652 &decoder->fixed_tl, &decoder->fixed_bl, NULL))
653 != 0)
654 {
655 decoder->fixed_tl = NULL;
656 return -1;
657 }
658
659 /* distance table */
660 for(i = 0; i < 30; i++) /* make an incomplete code set */
661 l[i] = 5;
662 decoder->fixed_bd = 5;
663 if((i = huft_build(l, 30, 0, cpdist, cpdext,
664 &decoder->fixed_td, &decoder->fixed_bd, NULL)) > 1)
665 {
666 huft_free(decoder->fixed_tl);
667 decoder->fixed_tl = NULL;
668 return -1;
669 }
670 }
671
672 decoder->tl = decoder->fixed_tl;
673 decoder->td = decoder->fixed_td;
674 decoder->bl = decoder->fixed_bl;
675 decoder->bd = decoder->fixed_bd;
676 return inflate_codes(decoder, buff, size);
677 }
678
inflate_dynamic(InflateHandler decoder,char * buff,long size)679 local long inflate_dynamic(InflateHandler decoder, char *buff, long size)
680 /* decompress an inflated type 2 (dynamic Huffman codes) block. */
681 {
682 int i; /* temporary variables */
683 unsigned j;
684 unsigned l; /* last length */
685 unsigned n; /* number of lengths to get */
686 struct huft *tl; /* literal/length code table */
687 struct huft *td; /* distance code table */
688 int bl; /* lookup bits for tl */
689 int bd; /* lookup bits for td */
690 unsigned nb; /* number of bit length codes */
691 unsigned nl; /* number of literal/length codes */
692 unsigned nd; /* number of distance codes */
693 #ifdef PKZIP_BUG_WORKAROUND
694 unsigned ll[288+32];/* literal/length and distance code lengths */
695 #else
696 unsigned ll[286+30];/* literal/length and distance code lengths */
697 #endif
698 static unsigned border[] = { /* Order of the bit length code lengths */
699 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15};
700 BITS_SAVE;
701
702 reuse_mblock(&decoder->pool);
703
704 /* read in table lengths */
705 NEEDBITS(5);
706 nl = 257 + GETBITS(5); /* number of literal/length codes */
707 DUMPBITS(5);
708 NEEDBITS(5);
709 nd = 1 + GETBITS(5); /* number of distance codes */
710 DUMPBITS(5);
711 NEEDBITS(4);
712 nb = 4 + GETBITS(4); /* number of bit length codes */
713 DUMPBITS(4);
714 #ifdef PKZIP_BUG_WORKAROUND
715 if(nl > 288 || nd > 32)
716 #else
717 if(nl > 286 || nd > 30)
718 #endif
719 {
720 BITS_RESTORE;
721 return -1; /* bad lengths */
722 }
723
724 /* read in bit-length-code lengths */
725 for(j = 0; j < nb; j++)
726 {
727 NEEDBITS(3);
728 ll[border[j]] = GETBITS(3);
729 DUMPBITS(3);
730 }
731 for(; j < 19; j++)
732 ll[border[j]] = 0;
733
734 /* build decoding table for trees--single level, 7 bit lookup */
735 bl = 7;
736 if((i = huft_build(ll, 19, 19, NULL, NULL, &tl, &bl, &decoder->pool)) != 0)
737 {
738 reuse_mblock(&decoder->pool);
739 BITS_RESTORE;
740 return -1; /* incomplete code set */
741 }
742
743 /* read in literal and distance code lengths */
744 n = nl + nd;
745 i = l = 0;
746 while((unsigned)i < n)
747 {
748 NEEDBITS((unsigned)bl);
749 j = (td = tl + (GETBITS(bl)))->b;
750 DUMPBITS(j);
751 j = td->v.n;
752 if(j < 16) /* length of code in bits (0..15) */
753 ll[i++] = l = j; /* save last length in l */
754 else if(j == 16) /* repeat last length 3 to 6 times */
755 {
756 NEEDBITS(2);
757 j = 3 + GETBITS(2);
758 DUMPBITS(2);
759 if((unsigned)i + j > n)
760 {
761 BITS_RESTORE;
762 return -1;
763 }
764 while(j--)
765 ll[i++] = l;
766 }
767 else if(j == 17) /* 3 to 10 zero length codes */
768 {
769 NEEDBITS(3);
770 j = 3 + GETBITS(3);
771 DUMPBITS(3);
772 if((unsigned)i + j > n)
773 {
774 BITS_RESTORE;
775 return -1;
776 }
777 while(j--)
778 ll[i++] = 0;
779 l = 0;
780 }
781 else /* j == 18: 11 to 138 zero length codes */
782 {
783 NEEDBITS(7);
784 j = 11 + GETBITS(7);
785 DUMPBITS(7);
786 if((unsigned)i + j > n)
787 {
788 BITS_RESTORE;
789 return -1;
790 }
791 while(j--)
792 ll[i++] = 0;
793 l = 0;
794 }
795 }
796
797 BITS_RESTORE;
798
799 /* free decoding table for trees */
800 reuse_mblock(&decoder->pool);
801
802 /* build the decoding tables for literal/length and distance codes */
803 bl = lbits;
804 i = huft_build(ll, nl, 257, cplens, cplext, &tl, &bl, &decoder->pool);
805 if(bl == 0) /* no literals or lengths */
806 i = 1;
807 if(i)
808 {
809 if(i == 1)
810 fprintf(stderr, " incomplete literal tree\n");
811 reuse_mblock(&decoder->pool);
812 return -1; /* incomplete code set */
813 }
814 bd = dbits;
815 i = huft_build(ll + nl, nd, 0, cpdist, cpdext, &td, &bd, &decoder->pool);
816 if(bd == 0 && nl > 257) /* lengths but no distances */
817 {
818 fprintf(stderr, " incomplete distance tree\n");
819 reuse_mblock(&decoder->pool);
820 return -1;
821 }
822
823 if(i == 1) {
824 #ifdef PKZIP_BUG_WORKAROUND
825 i = 0;
826 #else
827 fprintf(stderr, " incomplete distance tree\n");
828 #endif
829 }
830 if(i)
831 {
832 reuse_mblock(&decoder->pool);
833 return -1;
834 }
835
836 /* decompress until an end-of-block code */
837 decoder->tl = tl;
838 decoder->td = td;
839 decoder->bl = bl;
840 decoder->bd = bd;
841
842 i = inflate_codes(decoder, buff, size);
843
844 if(i == -1) /* error */
845 {
846 reuse_mblock(&decoder->pool);
847 return -1;
848 }
849
850 /* free the decoding tables, return */
851 return i;
852 }
853
inflate_start(InflateHandler decoder)854 local void inflate_start(InflateHandler decoder)
855 /* initialize window, bit buffer */
856 {
857 decoder->wp = 0;
858 decoder->bit_buf = 0;
859 decoder->bit_len = 0;
860 decoder->insize = decoder->inptr = 0;
861 decoder->fixed_td = decoder->fixed_tl = NULL;
862 decoder->method = -1;
863 decoder->eof = 0;
864 decoder->copy_leng = decoder->copy_dist = 0;
865 decoder->tl = NULL;
866
867 init_mblock(&decoder->pool);
868 }
869
870 /*ARGSUSED*/
default_read_func(char * buf,long size,void * v)871 static long default_read_func(char *buf, long size, void *v)
872 {
873 return (long)fread(buf, 1, size, stdin);
874 }
875
open_inflate_handler(long (* read_func)(char * buf,long size,void * user_val),void * user_val)876 InflateHandler open_inflate_handler(
877 long (* read_func)(char *buf, long size, void *user_val),
878 void *user_val)
879 {
880 InflateHandler decoder;
881
882 decoder = (InflateHandler)
883 malloc(sizeof(struct _InflateHandler));
884 inflate_start(decoder);
885 decoder->user_val = user_val;
886 if(read_func == NULL)
887 decoder->read_func = default_read_func;
888 else
889 decoder->read_func = read_func;
890 return decoder;
891 }
892
close_inflate_handler(InflateHandler decoder)893 void close_inflate_handler(InflateHandler decoder)
894 {
895 if(decoder->fixed_tl != NULL)
896 {
897 huft_free(decoder->fixed_td);
898 huft_free(decoder->fixed_tl);
899 decoder->fixed_td = decoder->fixed_tl = NULL;
900 }
901 reuse_mblock(&decoder->pool);
902 free(decoder);
903 }
904
905 /* decompress an inflated entry */
zip_inflate(InflateHandler decoder,char * buff,long size)906 long zip_inflate(
907 InflateHandler decoder,
908 char *buff,
909 long size)
910 {
911 long n, i;
912
913 n = 0;
914 while(n < size)
915 {
916 if(decoder->eof && decoder->method == -1)
917 return n;
918
919 if(decoder->copy_leng > 0)
920 {
921 unsigned l, w, d;
922
923 l = decoder->copy_leng;
924 w = decoder->wp;
925 if(decoder->method != STORED_BLOCK)
926 {
927 /* STATIC_TREES or DYN_TREES */
928 d = decoder->copy_dist;
929 while(l > 0 && n < size)
930 {
931 l--;
932 d &= WSIZE - 1;
933 w &= WSIZE - 1;
934 buff[n++] = decoder->slide[w++] = decoder->slide[d++];
935 }
936 decoder->copy_dist = d;
937 }
938 else /* STATIC_TREES or DYN_TREES */
939 {
940 BITS_SAVE;
941 while(l > 0 && n < size)
942 {
943 l--;
944 w &= WSIZE - 1;
945 NEEDBITS(8);
946 buff[n++] = decoder->slide[w++] = (uch)GETBITS(8);
947 DUMPBITS(8);
948 }
949 BITS_RESTORE;
950 if(l == 0)
951 decoder->method = -1; /* done */
952 }
953 decoder->copy_leng = l;
954 decoder->wp = w;
955 if(n == size)
956 return n;
957 }
958
959 if(decoder->method == -1)
960 {
961 BITS_SAVE;
962 if(decoder->eof)
963 {
964 BITS_RESTORE;
965 break;
966 }
967 /* read in last block bit */
968 NEEDBITS(1);
969 if(GETBITS(1))
970 decoder->eof = 1;
971 DUMPBITS(1);
972
973 /* read in block type */
974 NEEDBITS(2);
975 decoder->method = (int)GETBITS(2);
976 DUMPBITS(2);
977 decoder->tl = NULL;
978 decoder->copy_leng = 0;
979 BITS_RESTORE;
980 }
981
982 switch(decoder->method)
983 {
984 case STORED_BLOCK:
985 i = inflate_stored(decoder, buff + n, size - n);
986 break;
987
988 case STATIC_TREES:
989 if(decoder->tl != NULL)
990 i = inflate_codes(decoder, buff + n, size - n);
991 else
992 i = inflate_fixed(decoder, buff + n, size - n);
993 break;
994
995 case DYN_TREES:
996 if(decoder->tl != NULL)
997 i = inflate_codes(decoder, buff + n, size - n);
998 else
999 i = inflate_dynamic(decoder, buff + n, size - n);
1000 break;
1001
1002 default: /* error */
1003 i = -1;
1004 break;
1005 }
1006
1007 if(i == -1)
1008 {
1009 if(decoder->eof)
1010 return 0;
1011 return -1; /* error */
1012 }
1013 n += i;
1014 }
1015 return n;
1016 }
1017
1018 /* ===========================================================================
1019 * Fill the input buffer. This is called only when the buffer is empty.
1020 */
fill_inbuf(InflateHandler decoder)1021 local int fill_inbuf(InflateHandler decoder)
1022 {
1023 int len;
1024
1025 /* Read as much as possible */
1026 decoder->insize = 0;
1027 errno = 0;
1028 do {
1029 len = decoder->read_func((char*)decoder->inbuf + decoder->insize,
1030 (long)(INBUFSIZ - decoder->insize),
1031 decoder->user_val);
1032 if(len == 0 || len == EOF) break;
1033 decoder->insize += len;
1034 } while(decoder->insize < INBUFSIZ);
1035
1036 if(decoder->insize == 0)
1037 return EOF;
1038 decoder->inptr = 1;
1039 return decoder->inbuf[0];
1040 }
1041