Import 2.2.7pre1
[davej-history.git] / lib / inflate.c
blob011d382840b0bee098cbc1c0288f963dcbda7887
1 #define DEBG(x)
2 #define DEBG1(x)
3 /* inflate.c -- Not copyrighted 1992 by Mark Adler
4 version c10p1, 10 January 1993 */
6 /*
7 * Adapted for booting Linux by Hannu Savolainen 1993
8 * based on gzip-1.0.3
9 */
12 Inflate deflated (PKZIP's method 8 compressed) data. The compression
13 method searches for as much of the current string of bytes (up to a
14 length of 258) in the previous 32 K bytes. If it doesn't find any
15 matches (of at least length 3), it codes the next byte. Otherwise, it
16 codes the length of the matched string and its distance backwards from
17 the current position. There is a single Huffman code that codes both
18 single bytes (called "literals") and match lengths. A second Huffman
19 code codes the distance information, which follows a length code. Each
20 length or distance code actually represents a base value and a number
21 of "extra" (sometimes zero) bits to get to add to the base value. At
22 the end of each deflated block is a special end-of-block (EOB) literal/
23 length code. The decoding process is basically: get a literal/length
24 code; if EOB then done; if a literal, emit the decoded byte; if a
25 length then get the distance and emit the referred-to bytes from the
26 sliding window of previously emitted data.
28 There are (currently) three kinds of inflate blocks: stored, fixed, and
29 dynamic. The compressor deals with some chunk of data at a time, and
30 decides which method to use on a chunk-by-chunk basis. A chunk might
31 typically be 32 K or 64 K. If the chunk is incompressible, then the
32 "stored" method is used. In this case, the bytes are simply stored as
33 is, eight bits per byte, with none of the above coding. The bytes are
34 preceded by a count, since there is no longer an EOB code.
36 If the data is compressible, then either the fixed or dynamic methods
37 are used. In the dynamic method, the compressed data is preceded by
38 an encoding of the literal/length and distance Huffman codes that are
39 to be used to decode this block. The representation is itself Huffman
40 coded, and so is preceded by a description of that code. These code
41 descriptions take up a little space, and so for small blocks, there is
42 a predefined set of codes, called the fixed codes. The fixed method is
43 used if the block codes up smaller that way (usually for quite small
44 chunks), otherwise the dynamic method is used. In the latter case, the
45 codes are customized to the probabilities in the current block, and so
46 can code it much better than the pre-determined fixed codes.
48 The Huffman codes themselves are decoded using a multi-level table
49 lookup, in order to maximize the speed of decoding plus the speed of
50 building the decoding tables. See the comments below that precede the
51 lbits and dbits tuning parameters.
56 Notes beyond the 1.93a appnote.txt:
58 1. Distance pointers never point before the beginning of the output
59 stream.
60 2. Distance pointers can point back across blocks, up to 32k away.
61 3. There is an implied maximum of 7 bits for the bit length table and
62 15 bits for the actual data.
63 4. If only one code exists, then it is encoded using one bit. (Zero
64 would be more efficient, but perhaps a little confusing.) If two
65 codes exist, they are coded using one bit each (0 and 1).
66 5. There is no way of sending zero distance codes--a dummy must be
67 sent if there are none. (History: a pre 2.0 version of PKZIP would
68 store blocks with no distance codes, but this was discovered to be
69 too harsh a criterion.) Valid only for 1.93a. 2.04c does allow
70 zero distance codes, which is sent as one code of zero bits in
71 length.
72 6. There are up to 286 literal/length codes. Code 256 represents the
73 end-of-block. Note however that the static length tree defines
74 288 codes just to fill out the Huffman codes. Codes 286 and 287
75 cannot be used though, since there is no length base or extra bits
76 defined for them. Similarly, there are up to 30 distance codes.
77 However, static trees define 32 codes (all 5 bits) to fill out the
78 Huffman codes, but the last two had better not show up in the data.
79 7. Unzip can check dynamic Huffman blocks for complete code sets.
80 The exception is that a single code would not be complete (see #4).
81 8. The five bits following the block type is really the number of
82 literal codes sent minus 257.
83 9. Length codes 8,16,16 are interpreted as 13 length codes of 8 bits
84 (1+6+6). Therefore, to output three times the length, you output
85 three codes (1+1+1), whereas to output four times the same length,
86 you only need two codes (1+3). Hmm.
87 10. In the tree reconstruction algorithm, Code = Code + Increment
88 only if BitLength(i) is not zero. (Pretty obvious.)
89 11. Correction: 4 Bits: # of Bit Length codes - 4 (4 - 19)
90 12. Note: length code 284 can represent 227-258, but length code 285
91 really is 258. The last length deserves its own, short code
92 since it gets used a lot in very redundant files. The length
93 258 is special since 258 - 3 (the min match length) is 255.
94 13. The literal/length and distance code bit lengths are read as a
95 single stream of lengths. It is possible (and advantageous) for
96 a repeat code (16, 17, or 18) to go across the boundary between
97 the two sets of lengths.
100 #ifdef RCSID
101 static char rcsid[] = "#Id: inflate.c,v 0.14 1993/06/10 13:27:04 jloup Exp #";
102 #endif
104 #ifndef STATIC
106 #if defined(STDC_HEADERS) || defined(HAVE_STDLIB_H)
107 # include <sys/types.h>
108 # include <stdlib.h>
109 #endif
111 #include "gzip.h"
112 #define STATIC
113 #endif /* !STATIC */
115 #define slide window
117 /* Huffman code lookup table entry--this entry is four bytes for machines
118 that have 16-bit pointers (e.g. PC's in the small or medium model).
119 Valid extra bits are 0..13. e == 15 is EOB (end of block), e == 16
120 means that v is a literal, 16 < e < 32 means that v is a pointer to
121 the next table, which codes e - 16 bits, and lastly e == 99 indicates
122 an unused code. If a code with e == 99 is looked up, this implies an
123 error in the data. */
124 struct huft {
125 uch e; /* number of extra bits or operation */
126 uch b; /* number of bits in this code or subcode */
127 union {
128 ush n; /* literal, length base, or distance base */
129 struct huft *t; /* pointer to next level of table */
130 } v;
134 /* Function prototypes */
135 STATIC int huft_build OF((unsigned *, unsigned, unsigned, ush *, ush *,
136 struct huft **, int *));
137 STATIC int huft_free OF((struct huft *));
138 STATIC int inflate_codes OF((struct huft *, struct huft *, int, int));
139 STATIC int inflate_stored OF((void));
140 STATIC int inflate_fixed OF((void));
141 STATIC int inflate_dynamic OF((void));
142 STATIC int inflate_block OF((int *));
143 STATIC int inflate OF((void));
146 /* The inflate algorithm uses a sliding 32 K byte window on the uncompressed
147 stream to find repeated byte strings. This is implemented here as a
148 circular buffer. The index is updated simply by incrementing and then
149 ANDing with 0x7fff (32K-1). */
150 /* It is left to other modules to supply the 32 K area. It is assumed
151 to be usable as if it were declared "uch slide[32768];" or as just
152 "uch *slide;" and then malloc'ed in the latter case. The definition
153 must be in unzip.h, included above. */
154 /* unsigned wp; current position in slide */
155 #define wp outcnt
156 #define flush_output(w) (wp=(w),flush_window())
158 /* Tables for deflate from PKZIP's appnote.txt. */
159 static unsigned border[] = { /* Order of the bit length code lengths */
160 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15};
161 static ush cplens[] = { /* Copy lengths for literal codes 257..285 */
162 3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31,
163 35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, 227, 258, 0, 0};
164 /* note: see note #13 above about the 258 in this list. */
165 static ush cplext[] = { /* Extra bits for literal codes 257..285 */
166 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
167 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0, 99, 99}; /* 99==invalid */
168 static ush cpdist[] = { /* Copy offsets for distance codes 0..29 */
169 1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193,
170 257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, 6145,
171 8193, 12289, 16385, 24577};
172 static ush cpdext[] = { /* Extra bits for distance codes */
173 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
174 7, 7, 8, 8, 9, 9, 10, 10, 11, 11,
175 12, 12, 13, 13};
179 /* Macros for inflate() bit peeking and grabbing.
180 The usage is:
182 NEEDBITS(j)
183 x = b & mask_bits[j];
184 DUMPBITS(j)
186 where NEEDBITS makes sure that b has at least j bits in it, and
187 DUMPBITS removes the bits from b. The macros use the variable k
188 for the number of bits in b. Normally, b and k are register
189 variables for speed, and are initialized at the beginning of a
190 routine that uses these macros from a global bit buffer and count.
192 If we assume that EOB will be the longest code, then we will never
193 ask for bits with NEEDBITS that are beyond the end of the stream.
194 So, NEEDBITS should not read any more bytes than are needed to
195 meet the request. Then no bytes need to be "returned" to the buffer
196 at the end of the last block.
198 However, this assumption is not true for fixed blocks--the EOB code
199 is 7 bits, but the other literal/length codes can be 8 or 9 bits.
200 (The EOB code is shorter than other codes because fixed blocks are
201 generally short. So, while a block always has an EOB, many other
202 literal/length codes have a significantly lower probability of
203 showing up at all.) However, by making the first table have a
204 lookup of seven bits, the EOB code will be found in that first
205 lookup, and so will not require that too many bits be pulled from
206 the stream.
209 STATIC ulg bb; /* bit buffer */
210 STATIC unsigned bk; /* bits in bit buffer */
212 STATIC ush mask_bits[] = {
213 0x0000,
214 0x0001, 0x0003, 0x0007, 0x000f, 0x001f, 0x003f, 0x007f, 0x00ff,
215 0x01ff, 0x03ff, 0x07ff, 0x0fff, 0x1fff, 0x3fff, 0x7fff, 0xffff
218 #define NEXTBYTE() (uch)get_byte()
219 #define NEEDBITS(n) {while(k<(n)){b|=((ulg)NEXTBYTE())<<k;k+=8;}}
220 #define DUMPBITS(n) {b>>=(n);k-=(n);}
224 Huffman code decoding is performed using a multi-level table lookup.
225 The fastest way to decode is to simply build a lookup table whose
226 size is determined by the longest code. However, the time it takes
227 to build this table can also be a factor if the data being decoded
228 is not very long. The most common codes are necessarily the
229 shortest codes, so those codes dominate the decoding time, and hence
230 the speed. The idea is you can have a shorter table that decodes the
231 shorter, more probable codes, and then point to subsidiary tables for
232 the longer codes. The time it costs to decode the longer codes is
233 then traded against the time it takes to make longer tables.
235 This results of this trade are in the variables lbits and dbits
236 below. lbits is the number of bits the first level table for literal/
237 length codes can decode in one step, and dbits is the same thing for
238 the distance codes. Subsequent tables are also less than or equal to
239 those sizes. These values may be adjusted either when all of the
240 codes are shorter than that, in which case the longest code length in
241 bits is used, or when the shortest code is *longer* than the requested
242 table size, in which case the length of the shortest code in bits is
243 used.
245 There are two different values for the two tables, since they code a
246 different number of possibilities each. The literal/length table
247 codes 286 possible values, or in a flat code, a little over eight
248 bits. The distance table codes 30 possible values, or a little less
249 than five bits, flat. The optimum values for speed end up being
250 about one bit more than those, so lbits is 8+1 and dbits is 5+1.
251 The optimum values may differ though from machine to machine, and
252 possibly even between compilers. Your mileage may vary.
256 STATIC int lbits = 9; /* bits in base literal/length lookup table */
257 STATIC int dbits = 6; /* bits in base distance lookup table */
260 /* If BMAX needs to be larger than 16, then h and x[] should be ulg. */
261 #define BMAX 16 /* maximum bit length of any code (16 for explode) */
262 #define N_MAX 288 /* maximum number of codes in any set */
265 STATIC unsigned hufts; /* track memory usage */
268 STATIC int huft_build(b, n, s, d, e, t, m)
269 unsigned *b; /* code lengths in bits (all assumed <= BMAX) */
270 unsigned n; /* number of codes (assumed <= N_MAX) */
271 unsigned s; /* number of simple-valued codes (0..s-1) */
272 ush *d; /* list of base values for non-simple codes */
273 ush *e; /* list of extra bits for non-simple codes */
274 struct huft **t; /* result: starting table */
275 int *m; /* maximum lookup bits, returns actual */
276 /* Given a list of code lengths and a maximum table size, make a set of
277 tables to decode that set of codes. Return zero on success, one if
278 the given code set is incomplete (the tables are still built in this
279 case), two if the input is invalid (all zero length codes or an
280 oversubscribed set of lengths), and three if not enough memory. */
282 unsigned a; /* counter for codes of length k */
283 unsigned c[BMAX+1]; /* bit length count table */
284 unsigned f; /* i repeats in table every f entries */
285 int g; /* maximum code length */
286 int h; /* table level */
287 register unsigned i; /* counter, current code */
288 register unsigned j; /* counter */
289 register int k; /* number of bits in current code */
290 int l; /* bits per table (returned in m) */
291 register unsigned *p; /* pointer into c[], b[], or v[] */
292 register struct huft *q; /* points to current table */
293 struct huft r; /* table entry for structure assignment */
294 struct huft *u[BMAX]; /* table stack */
295 unsigned v[N_MAX]; /* values in order of bit length */
296 register int w; /* bits before this table == (l * h) */
297 unsigned x[BMAX+1]; /* bit offsets, then code stack */
298 unsigned *xp; /* pointer into x */
299 int y; /* number of dummy codes added */
300 unsigned z; /* number of entries in current table */
302 DEBG("huft1 ");
304 /* Generate counts for each bit length */
305 memzero(c, sizeof(c));
306 p = b; i = n;
307 do {
308 Tracecv(*p, (stderr, (n-i >= ' ' && n-i <= '~' ? "%c %d\n" : "0x%x %d\n"),
309 n-i, *p));
310 c[*p]++; /* assume all entries <= BMAX */
311 p++; /* Can't combine with above line (Solaris bug) */
312 } while (--i);
313 if (c[0] == n) /* null input--all zero length codes */
315 *t = (struct huft *)NULL;
316 *m = 0;
317 return 0;
320 DEBG("huft2 ");
322 /* Find minimum and maximum length, bound *m by those */
323 l = *m;
324 for (j = 1; j <= BMAX; j++)
325 if (c[j])
326 break;
327 k = j; /* minimum code length */
328 if ((unsigned)l < j)
329 l = j;
330 for (i = BMAX; i; i--)
331 if (c[i])
332 break;
333 g = i; /* maximum code length */
334 if ((unsigned)l > i)
335 l = i;
336 *m = l;
338 DEBG("huft3 ");
340 /* Adjust last length count to fill out codes, if needed */
341 for (y = 1 << j; j < i; j++, y <<= 1)
342 if ((y -= c[j]) < 0)
343 return 2; /* bad input: more codes than bits */
344 if ((y -= c[i]) < 0)
345 return 2;
346 c[i] += y;
348 DEBG("huft4 ");
350 /* Generate starting offsets into the value table for each length */
351 x[1] = j = 0;
352 p = c + 1; xp = x + 2;
353 while (--i) { /* note that i == g from above */
354 *xp++ = (j += *p++);
357 DEBG("huft5 ");
359 /* Make a table of values in order of bit lengths */
360 p = b; i = 0;
361 do {
362 if ((j = *p++) != 0)
363 v[x[j]++] = i;
364 } while (++i < n);
366 DEBG("h6 ");
368 /* Generate the Huffman codes and for each, make the table entries */
369 x[0] = i = 0; /* first Huffman code is zero */
370 p = v; /* grab values in bit order */
371 h = -1; /* no tables yet--level -1 */
372 w = -l; /* bits decoded == (l * h) */
373 u[0] = (struct huft *)NULL; /* just to keep compilers happy */
374 q = (struct huft *)NULL; /* ditto */
375 z = 0; /* ditto */
376 DEBG("h6a ");
378 /* go through the bit lengths (k already is bits in shortest code) */
379 for (; k <= g; k++)
381 DEBG("h6b ");
382 a = c[k];
383 while (a--)
385 DEBG("h6b1 ");
386 /* here i is the Huffman code of length k bits for value *p */
387 /* make tables up to required level */
388 while (k > w + l)
390 DEBG1("1 ");
391 h++;
392 w += l; /* previous table always l bits */
394 /* compute minimum size table less than or equal to l bits */
395 z = (z = g - w) > (unsigned)l ? l : z; /* upper limit on table size */
396 if ((f = 1 << (j = k - w)) > a + 1) /* try a k-w bit table */
397 { /* too few codes for k-w bit table */
398 DEBG1("2 ");
399 f -= a + 1; /* deduct codes from patterns left */
400 xp = c + k;
401 while (++j < z) /* try smaller tables up to z bits */
403 if ((f <<= 1) <= *++xp)
404 break; /* enough codes to use up j bits */
405 f -= *xp; /* else deduct codes from patterns */
408 DEBG1("3 ");
409 z = 1 << j; /* table entries for j-bit table */
411 /* allocate and link in new table */
412 if ((q = (struct huft *)malloc((z + 1)*sizeof(struct huft))) ==
413 (struct huft *)NULL)
415 if (h)
416 huft_free(u[0]);
417 return 3; /* not enough memory */
419 DEBG1("4 ");
420 hufts += z + 1; /* track memory usage */
421 *t = q + 1; /* link to list for huft_free() */
422 *(t = &(q->v.t)) = (struct huft *)NULL;
423 u[h] = ++q; /* table starts after link */
425 DEBG1("5 ");
426 /* connect to last table, if there is one */
427 if (h)
429 x[h] = i; /* save pattern for backing up */
430 r.b = (uch)l; /* bits to dump before this table */
431 r.e = (uch)(16 + j); /* bits in this table */
432 r.v.t = q; /* pointer to this table */
433 j = i >> (w - l); /* (get around Turbo C bug) */
434 u[h-1][j] = r; /* connect to last table */
436 DEBG1("6 ");
438 DEBG("h6c ");
440 /* set up table entry in r */
441 r.b = (uch)(k - w);
442 if (p >= v + n)
443 r.e = 99; /* out of values--invalid code */
444 else if (*p < s)
446 r.e = (uch)(*p < 256 ? 16 : 15); /* 256 is end-of-block code */
447 r.v.n = (ush)(*p); /* simple code is just the value */
448 p++; /* one compiler does not like *p++ */
450 else
452 r.e = (uch)e[*p - s]; /* non-simple--look up in lists */
453 r.v.n = d[*p++ - s];
455 DEBG("h6d ");
457 /* fill code-like entries with r */
458 f = 1 << (k - w);
459 for (j = i >> w; j < z; j += f)
460 q[j] = r;
462 /* backwards increment the k-bit code i */
463 for (j = 1 << (k - 1); i & j; j >>= 1)
464 i ^= j;
465 i ^= j;
467 /* backup over finished tables */
468 while ((i & ((1 << w) - 1)) != x[h])
470 h--; /* don't need to update q */
471 w -= l;
473 DEBG("h6e ");
475 DEBG("h6f ");
478 DEBG("huft7 ");
480 /* Return true (1) if we were given an incomplete table */
481 return y != 0 && g != 1;
486 STATIC int huft_free(t)
487 struct huft *t; /* table to free */
488 /* Free the malloc'ed tables built by huft_build(), which makes a linked
489 list of the tables it made, with the links in a dummy first entry of
490 each table. */
492 register struct huft *p, *q;
495 /* Go through linked list, freeing from the malloced (t[-1]) address. */
496 p = t;
497 while (p != (struct huft *)NULL)
499 q = (--p)->v.t;
500 free((char*)p);
501 p = q;
503 return 0;
507 STATIC int inflate_codes(tl, td, bl, bd)
508 struct huft *tl, *td; /* literal/length and distance decoder tables */
509 int bl, bd; /* number of bits decoded by tl[] and td[] */
510 /* inflate (decompress) the codes in a deflated (compressed) block.
511 Return an error code or zero if it all goes ok. */
513 register unsigned e; /* table entry flag/number of extra bits */
514 unsigned n, d; /* length and index for copy */
515 unsigned w; /* current window position */
516 struct huft *t; /* pointer to table entry */
517 unsigned ml, md; /* masks for bl and bd bits */
518 register ulg b; /* bit buffer */
519 register unsigned k; /* number of bits in bit buffer */
522 /* make local copies of globals */
523 b = bb; /* initialize bit buffer */
524 k = bk;
525 w = wp; /* initialize window position */
527 /* inflate the coded data */
528 ml = mask_bits[bl]; /* precompute masks for speed */
529 md = mask_bits[bd];
530 for (;;) /* do until end of block */
532 NEEDBITS((unsigned)bl)
533 if ((e = (t = tl + ((unsigned)b & ml))->e) > 16)
534 do {
535 if (e == 99)
536 return 1;
537 DUMPBITS(t->b)
538 e -= 16;
539 NEEDBITS(e)
540 } while ((e = (t = t->v.t + ((unsigned)b & mask_bits[e]))->e) > 16);
541 DUMPBITS(t->b)
542 if (e == 16) /* then it's a literal */
544 slide[w++] = (uch)t->v.n;
545 Tracevv((stderr, "%c", slide[w-1]));
546 if (w == WSIZE)
548 flush_output(w);
549 w = 0;
552 else /* it's an EOB or a length */
554 /* exit if end of block */
555 if (e == 15)
556 break;
558 /* get length of block to copy */
559 NEEDBITS(e)
560 n = t->v.n + ((unsigned)b & mask_bits[e]);
561 DUMPBITS(e);
563 /* decode distance of block to copy */
564 NEEDBITS((unsigned)bd)
565 if ((e = (t = td + ((unsigned)b & md))->e) > 16)
566 do {
567 if (e == 99)
568 return 1;
569 DUMPBITS(t->b)
570 e -= 16;
571 NEEDBITS(e)
572 } while ((e = (t = t->v.t + ((unsigned)b & mask_bits[e]))->e) > 16);
573 DUMPBITS(t->b)
574 NEEDBITS(e)
575 d = w - t->v.n - ((unsigned)b & mask_bits[e]);
576 DUMPBITS(e)
577 Tracevv((stderr,"\\[%d,%d]", w-d, n));
579 /* do the copy */
580 do {
581 n -= (e = (e = WSIZE - ((d &= WSIZE-1) > w ? d : w)) > n ? n : e);
582 #if !defined(NOMEMCPY) && !defined(DEBUG)
583 if (w - d >= e) /* (this test assumes unsigned comparison) */
585 memcpy(slide + w, slide + d, e);
586 w += e;
587 d += e;
589 else /* do it slow to avoid memcpy() overlap */
590 #endif /* !NOMEMCPY */
591 do {
592 slide[w++] = slide[d++];
593 Tracevv((stderr, "%c", slide[w-1]));
594 } while (--e);
595 if (w == WSIZE)
597 flush_output(w);
598 w = 0;
600 } while (n);
605 /* restore the globals from the locals */
606 wp = w; /* restore global window pointer */
607 bb = b; /* restore global bit buffer */
608 bk = k;
610 /* done */
611 return 0;
616 STATIC int inflate_stored()
617 /* "decompress" an inflated type 0 (stored) block. */
619 unsigned n; /* number of bytes in block */
620 unsigned w; /* current window position */
621 register ulg b; /* bit buffer */
622 register unsigned k; /* number of bits in bit buffer */
624 DEBG("<stor");
626 /* make local copies of globals */
627 b = bb; /* initialize bit buffer */
628 k = bk;
629 w = wp; /* initialize window position */
632 /* go to byte boundary */
633 n = k & 7;
634 DUMPBITS(n);
637 /* get the length and its complement */
638 NEEDBITS(16)
639 n = ((unsigned)b & 0xffff);
640 DUMPBITS(16)
641 NEEDBITS(16)
642 if (n != (unsigned)((~b) & 0xffff))
643 return 1; /* error in compressed data */
644 DUMPBITS(16)
647 /* read and output the compressed data */
648 while (n--)
650 NEEDBITS(8)
651 slide[w++] = (uch)b;
652 if (w == WSIZE)
654 flush_output(w);
655 w = 0;
657 DUMPBITS(8)
661 /* restore the globals from the locals */
662 wp = w; /* restore global window pointer */
663 bb = b; /* restore global bit buffer */
664 bk = k;
666 DEBG(">");
667 return 0;
672 STATIC int inflate_fixed()
673 /* decompress an inflated type 1 (fixed Huffman codes) block. We should
674 either replace this with a custom decoder, or at least precompute the
675 Huffman tables. */
677 int i; /* temporary variable */
678 struct huft *tl; /* literal/length code table */
679 struct huft *td; /* distance code table */
680 int bl; /* lookup bits for tl */
681 int bd; /* lookup bits for td */
682 unsigned l[288]; /* length list for huft_build */
684 DEBG("<fix");
686 /* set up literal table */
687 for (i = 0; i < 144; i++)
688 l[i] = 8;
689 for (; i < 256; i++)
690 l[i] = 9;
691 for (; i < 280; i++)
692 l[i] = 7;
693 for (; i < 288; i++) /* make a complete, but wrong code set */
694 l[i] = 8;
695 bl = 7;
696 if ((i = huft_build(l, 288, 257, cplens, cplext, &tl, &bl)) != 0)
697 return i;
700 /* set up distance table */
701 for (i = 0; i < 30; i++) /* make an incomplete code set */
702 l[i] = 5;
703 bd = 5;
704 if ((i = huft_build(l, 30, 0, cpdist, cpdext, &td, &bd)) > 1)
706 huft_free(tl);
708 DEBG(">");
709 return i;
713 /* decompress until an end-of-block code */
714 if (inflate_codes(tl, td, bl, bd))
715 return 1;
718 /* free the decoding tables, return */
719 huft_free(tl);
720 huft_free(td);
721 return 0;
726 STATIC int inflate_dynamic()
727 /* decompress an inflated type 2 (dynamic Huffman codes) block. */
729 int i; /* temporary variables */
730 unsigned j;
731 unsigned l; /* last length */
732 unsigned m; /* mask for bit lengths table */
733 unsigned n; /* number of lengths to get */
734 struct huft *tl; /* literal/length code table */
735 struct huft *td; /* distance code table */
736 int bl; /* lookup bits for tl */
737 int bd; /* lookup bits for td */
738 unsigned nb; /* number of bit length codes */
739 unsigned nl; /* number of literal/length codes */
740 unsigned nd; /* number of distance codes */
741 #ifdef PKZIP_BUG_WORKAROUND
742 unsigned ll[288+32]; /* literal/length and distance code lengths */
743 #else
744 unsigned ll[286+30]; /* literal/length and distance code lengths */
745 #endif
746 register ulg b; /* bit buffer */
747 register unsigned k; /* number of bits in bit buffer */
749 DEBG("<dyn");
751 /* make local bit buffer */
752 b = bb;
753 k = bk;
756 /* read in table lengths */
757 NEEDBITS(5)
758 nl = 257 + ((unsigned)b & 0x1f); /* number of literal/length codes */
759 DUMPBITS(5)
760 NEEDBITS(5)
761 nd = 1 + ((unsigned)b & 0x1f); /* number of distance codes */
762 DUMPBITS(5)
763 NEEDBITS(4)
764 nb = 4 + ((unsigned)b & 0xf); /* number of bit length codes */
765 DUMPBITS(4)
766 #ifdef PKZIP_BUG_WORKAROUND
767 if (nl > 288 || nd > 32)
768 #else
769 if (nl > 286 || nd > 30)
770 #endif
771 return 1; /* bad lengths */
773 DEBG("dyn1 ");
775 /* read in bit-length-code lengths */
776 for (j = 0; j < nb; j++)
778 NEEDBITS(3)
779 ll[border[j]] = (unsigned)b & 7;
780 DUMPBITS(3)
782 for (; j < 19; j++)
783 ll[border[j]] = 0;
785 DEBG("dyn2 ");
787 /* build decoding table for trees--single level, 7 bit lookup */
788 bl = 7;
789 if ((i = huft_build(ll, 19, 19, NULL, NULL, &tl, &bl)) != 0)
791 if (i == 1)
792 huft_free(tl);
793 return i; /* incomplete code set */
796 DEBG("dyn3 ");
798 /* read in literal and distance code lengths */
799 n = nl + nd;
800 m = mask_bits[bl];
801 i = l = 0;
802 while ((unsigned)i < n)
804 NEEDBITS((unsigned)bl)
805 j = (td = tl + ((unsigned)b & m))->b;
806 DUMPBITS(j)
807 j = td->v.n;
808 if (j < 16) /* length of code in bits (0..15) */
809 ll[i++] = l = j; /* save last length in l */
810 else if (j == 16) /* repeat last length 3 to 6 times */
812 NEEDBITS(2)
813 j = 3 + ((unsigned)b & 3);
814 DUMPBITS(2)
815 if ((unsigned)i + j > n)
816 return 1;
817 while (j--)
818 ll[i++] = l;
820 else if (j == 17) /* 3 to 10 zero length codes */
822 NEEDBITS(3)
823 j = 3 + ((unsigned)b & 7);
824 DUMPBITS(3)
825 if ((unsigned)i + j > n)
826 return 1;
827 while (j--)
828 ll[i++] = 0;
829 l = 0;
831 else /* j == 18: 11 to 138 zero length codes */
833 NEEDBITS(7)
834 j = 11 + ((unsigned)b & 0x7f);
835 DUMPBITS(7)
836 if ((unsigned)i + j > n)
837 return 1;
838 while (j--)
839 ll[i++] = 0;
840 l = 0;
844 DEBG("dyn4 ");
846 /* free decoding table for trees */
847 huft_free(tl);
849 DEBG("dyn5 ");
851 /* restore the global bit buffer */
852 bb = b;
853 bk = k;
855 DEBG("dyn5a ");
857 /* build the decoding tables for literal/length and distance codes */
858 bl = lbits;
859 if ((i = huft_build(ll, nl, 257, cplens, cplext, &tl, &bl)) != 0)
861 DEBG("dyn5b ");
862 if (i == 1) {
863 error(" incomplete literal tree\n");
864 huft_free(tl);
866 return i; /* incomplete code set */
868 DEBG("dyn5c ");
869 bd = dbits;
870 if ((i = huft_build(ll + nl, nd, 0, cpdist, cpdext, &td, &bd)) != 0)
872 DEBG("dyn5d ");
873 if (i == 1) {
874 error(" incomplete distance tree\n");
875 #ifdef PKZIP_BUG_WORKAROUND
876 i = 0;
878 #else
879 huft_free(td);
881 huft_free(tl);
882 return i; /* incomplete code set */
883 #endif
886 DEBG("dyn6 ");
888 /* decompress until an end-of-block code */
889 if (inflate_codes(tl, td, bl, bd))
890 return 1;
892 DEBG("dyn7 ");
894 /* free the decoding tables, return */
895 huft_free(tl);
896 huft_free(td);
898 DEBG(">");
899 return 0;
904 STATIC int inflate_block(e)
905 int *e; /* last block flag */
906 /* decompress an inflated block */
908 unsigned t; /* block type */
909 register ulg b; /* bit buffer */
910 register unsigned k; /* number of bits in bit buffer */
912 DEBG("<blk");
914 /* make local bit buffer */
915 b = bb;
916 k = bk;
919 /* read in last block bit */
920 NEEDBITS(1)
921 *e = (int)b & 1;
922 DUMPBITS(1)
925 /* read in block type */
926 NEEDBITS(2)
927 t = (unsigned)b & 3;
928 DUMPBITS(2)
931 /* restore the global bit buffer */
932 bb = b;
933 bk = k;
935 /* inflate that block type */
936 if (t == 2)
937 return inflate_dynamic();
938 if (t == 0)
939 return inflate_stored();
940 if (t == 1)
941 return inflate_fixed();
943 DEBG(">");
945 /* bad block type */
946 return 2;
951 STATIC int inflate()
952 /* decompress an inflated entry */
954 int e; /* last block flag */
955 int r; /* result code */
956 unsigned h; /* maximum struct huft's malloc'ed */
957 void *ptr;
959 /* initialize window, bit buffer */
960 wp = 0;
961 bk = 0;
962 bb = 0;
965 /* decompress until the last block */
966 h = 0;
967 do {
968 hufts = 0;
969 gzip_mark(&ptr);
970 if ((r = inflate_block(&e)) != 0) {
971 gzip_release(&ptr);
972 return r;
974 gzip_release(&ptr);
975 if (hufts > h)
976 h = hufts;
977 } while (!e);
979 /* Undo too much lookahead. The next read will be byte aligned so we
980 * can discard unused bits in the last meaningful byte.
982 while (bk >= 8) {
983 bk -= 8;
984 inptr--;
987 /* flush out slide */
988 flush_output(wp);
991 /* return success */
992 #ifdef DEBUG
993 fprintf(stderr, "<%u> ", h);
994 #endif /* DEBUG */
995 return 0;
998 /**********************************************************************
1000 * The following are support routines for inflate.c
1002 **********************************************************************/
1004 static ulg crc_32_tab[256];
1005 static ulg crc = (ulg)0xffffffffL; /* shift register contents */
1006 #define CRC_VALUE (crc ^ 0xffffffffL)
1009 * Code to compute the CRC-32 table. Borrowed from
1010 * gzip-1.0.3/makecrc.c.
1013 static void
1014 makecrc(void)
1016 /* Not copyrighted 1990 Mark Adler */
1018 unsigned long c; /* crc shift register */
1019 unsigned long e; /* polynomial exclusive-or pattern */
1020 int i; /* counter for all possible eight bit values */
1021 int k; /* byte being shifted into crc apparatus */
1023 /* terms of polynomial defining this crc (except x^32): */
1024 static int p[] = {0,1,2,4,5,7,8,10,11,12,16,22,23,26};
1026 /* Make exclusive-or pattern from polynomial */
1027 e = 0;
1028 for (i = 0; i < sizeof(p)/sizeof(int); i++)
1029 e |= 1L << (31 - p[i]);
1031 crc_32_tab[0] = 0;
1033 for (i = 1; i < 256; i++)
1035 c = 0;
1036 for (k = i | 256; k != 1; k >>= 1)
1038 c = c & 1 ? (c >> 1) ^ e : c >> 1;
1039 if (k & 1)
1040 c ^= e;
1042 crc_32_tab[i] = c;
1046 /* gzip flag byte */
1047 #define ASCII_FLAG 0x01 /* bit 0 set: file probably ASCII text */
1048 #define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */
1049 #define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */
1050 #define ORIG_NAME 0x08 /* bit 3 set: original file name present */
1051 #define COMMENT 0x10 /* bit 4 set: file comment present */
1052 #define ENCRYPTED 0x20 /* bit 5 set: file is encrypted */
1053 #define RESERVED 0xC0 /* bit 6,7: reserved */
1056 * Do the uncompression!
1058 static int gunzip(void)
1060 uch flags;
1061 unsigned char magic[2]; /* magic header */
1062 char method;
1063 ulg orig_crc = 0; /* original crc */
1064 ulg orig_len = 0; /* original uncompressed length */
1065 int res;
1067 magic[0] = (unsigned char)get_byte();
1068 magic[1] = (unsigned char)get_byte();
1069 method = (unsigned char)get_byte();
1071 if (magic[0] != 037 ||
1072 ((magic[1] != 0213) && (magic[1] != 0236))) {
1073 error("bad gzip magic numbers");
1074 return -1;
1077 /* We only support method #8, DEFLATED */
1078 if (method != 8) {
1079 error("internal error, invalid method");
1080 return -1;
1083 flags = (uch)get_byte();
1084 if ((flags & ENCRYPTED) != 0) {
1085 error("Input is encrypted\n");
1086 return -1;
1088 if ((flags & CONTINUATION) != 0) {
1089 error("Multi part input\n");
1090 return -1;
1092 if ((flags & RESERVED) != 0) {
1093 error("Input has invalid flags\n");
1094 return -1;
1096 (ulg)get_byte(); /* Get timestamp */
1097 ((ulg)get_byte()) << 8;
1098 ((ulg)get_byte()) << 16;
1099 ((ulg)get_byte()) << 24;
1101 (void)get_byte(); /* Ignore extra flags for the moment */
1102 (void)get_byte(); /* Ignore OS type for the moment */
1104 if ((flags & EXTRA_FIELD) != 0) {
1105 unsigned len = (unsigned)get_byte();
1106 len |= ((unsigned)get_byte())<<8;
1107 while (len--) (void)get_byte();
1110 /* Get original file name if it was truncated */
1111 if ((flags & ORIG_NAME) != 0) {
1112 /* Discard the old name */
1113 while (get_byte() != 0) /* null */ ;
1116 /* Discard file comment if any */
1117 if ((flags & COMMENT) != 0) {
1118 while (get_byte() != 0) /* null */ ;
1121 /* Decompress */
1122 if ((res = inflate())) {
1123 switch (res) {
1124 case 0:
1125 break;
1126 case 1:
1127 error("invalid compressed format (err=1)");
1128 break;
1129 case 2:
1130 error("invalid compressed format (err=2)");
1131 break;
1132 case 3:
1133 error("out of memory");
1134 break;
1135 default:
1136 error("invalid compressed format (other)");
1138 return -1;
1141 /* Get the crc and original length */
1142 /* crc32 (see algorithm.doc)
1143 * uncompressed input size modulo 2^32
1145 orig_crc = (ulg) get_byte();
1146 orig_crc |= (ulg) get_byte() << 8;
1147 orig_crc |= (ulg) get_byte() << 16;
1148 orig_crc |= (ulg) get_byte() << 24;
1150 orig_len = (ulg) get_byte();
1151 orig_len |= (ulg) get_byte() << 8;
1152 orig_len |= (ulg) get_byte() << 16;
1153 orig_len |= (ulg) get_byte() << 24;
1155 /* Validate decompression */
1156 if (orig_crc != CRC_VALUE) {
1157 error("crc error");
1158 return -1;
1160 if (orig_len != bytes_out) {
1161 error("length error");
1162 return -1;
1164 return 0;