9e052883 |
1 | /* blast.c |
2 | * Copyright (C) 2003, 2012, 2013 Mark Adler |
3 | * For conditions of distribution and use, see copyright notice in blast.h |
4 | * version 1.3, 24 Aug 2013 |
5 | * |
6 | * blast.c decompresses data compressed by the PKWare Compression Library. |
7 | * This function provides functionality similar to the explode() function of |
8 | * the PKWare library, hence the name "blast". |
9 | * |
10 | * This decompressor is based on the excellent format description provided by |
11 | * Ben Rudiak-Gould in comp.compression on August 13, 2001. Interestingly, the |
12 | * example Ben provided in the post is incorrect. The distance 110001 should |
13 | * instead be 111000. When corrected, the example byte stream becomes: |
14 | * |
15 | * 00 04 82 24 25 8f 80 7f |
16 | * |
17 | * which decompresses to "AIAIAIAIAIAIA" (without the quotes). |
18 | */ |
19 | |
20 | /* |
21 | * Change history: |
22 | * |
23 | * 1.0 12 Feb 2003 - First version |
24 | * 1.1 16 Feb 2003 - Fixed distance check for > 4 GB uncompressed data |
25 | * 1.2 24 Oct 2012 - Add note about using binary mode in stdio |
26 | * - Fix comparisons of differently signed integers |
27 | * 1.3 24 Aug 2013 - Return unused input from blast() |
28 | * - Fix test code to correctly report unused input |
29 | * - Enable the provision of initial input to blast() |
30 | */ |
31 | |
32 | #include <stddef.h> /* for NULL */ |
33 | #include <setjmp.h> /* for setjmp(), longjmp(), and jmp_buf */ |
34 | #include "blast.h" /* prototype for blast() */ |
35 | |
36 | #define local static /* for local function definitions */ |
37 | #define MAXBITS 13 /* maximum code length */ |
38 | #define MAXWIN 4096 /* maximum window size */ |
39 | |
40 | /* input and output state */ |
41 | struct state { |
42 | /* input state */ |
43 | blast_in infun; /* input function provided by user */ |
44 | void *inhow; /* opaque information passed to infun() */ |
45 | unsigned char *in; /* next input location */ |
46 | unsigned left; /* available input at in */ |
47 | int bitbuf; /* bit buffer */ |
48 | int bitcnt; /* number of bits in bit buffer */ |
49 | |
50 | /* input limit error return state for bits() and decode() */ |
51 | jmp_buf env; |
52 | |
53 | /* output state */ |
54 | blast_out outfun; /* output function provided by user */ |
55 | void *outhow; /* opaque information passed to outfun() */ |
56 | unsigned next; /* index of next write location in out[] */ |
57 | int first; /* true to check distances (for first 4K) */ |
58 | unsigned char out[MAXWIN]; /* output buffer and sliding window */ |
59 | }; |
60 | |
61 | /* |
62 | * Return need bits from the input stream. This always leaves less than |
63 | * eight bits in the buffer. bits() works properly for need == 0. |
64 | * |
65 | * Format notes: |
66 | * |
67 | * - Bits are stored in bytes from the least significant bit to the most |
68 | * significant bit. Therefore bits are dropped from the bottom of the bit |
69 | * buffer, using shift right, and new bytes are appended to the top of the |
70 | * bit buffer, using shift left. |
71 | */ |
72 | local int bits(struct state *s, int need) |
73 | { |
74 | int val; /* bit accumulator */ |
75 | |
76 | /* load at least need bits into val */ |
77 | val = s->bitbuf; |
78 | while (s->bitcnt < need) { |
79 | if (s->left == 0) { |
80 | s->left = s->infun(s->inhow, &(s->in)); |
81 | if (s->left == 0) longjmp(s->env, 1); /* out of input */ |
82 | } |
83 | val |= (int)(*(s->in)++) << s->bitcnt; /* load eight bits */ |
84 | s->left--; |
85 | s->bitcnt += 8; |
86 | } |
87 | |
88 | /* drop need bits and update buffer, always zero to seven bits left */ |
89 | s->bitbuf = val >> need; |
90 | s->bitcnt -= need; |
91 | |
92 | /* return need bits, zeroing the bits above that */ |
93 | return val & ((1 << need) - 1); |
94 | } |
95 | |
96 | /* |
97 | * Huffman code decoding tables. count[1..MAXBITS] is the number of symbols of |
98 | * each length, which for a canonical code are stepped through in order. |
99 | * symbol[] are the symbol values in canonical order, where the number of |
100 | * entries is the sum of the counts in count[]. The decoding process can be |
101 | * seen in the function decode() below. |
102 | */ |
103 | struct huffman { |
104 | short *count; /* number of symbols of each length */ |
105 | short *symbol; /* canonically ordered symbols */ |
106 | }; |
107 | |
108 | /* |
109 | * Decode a code from the stream s using huffman table h. Return the symbol or |
110 | * a negative value if there is an error. If all of the lengths are zero, i.e. |
111 | * an empty code, or if the code is incomplete and an invalid code is received, |
112 | * then -9 is returned after reading MAXBITS bits. |
113 | * |
114 | * Format notes: |
115 | * |
116 | * - The codes as stored in the compressed data are bit-reversed relative to |
117 | * a simple integer ordering of codes of the same lengths. Hence below the |
118 | * bits are pulled from the compressed data one at a time and used to |
119 | * build the code value reversed from what is in the stream in order to |
120 | * permit simple integer comparisons for decoding. |
121 | * |
122 | * - The first code for the shortest length is all ones. Subsequent codes of |
123 | * the same length are simply integer decrements of the previous code. When |
124 | * moving up a length, a one bit is appended to the code. For a complete |
125 | * code, the last code of the longest length will be all zeros. To support |
126 | * this ordering, the bits pulled during decoding are inverted to apply the |
127 | * more "natural" ordering starting with all zeros and incrementing. |
128 | */ |
129 | local int decode(struct state *s, struct huffman *h) |
130 | { |
131 | int len; /* current number of bits in code */ |
132 | int code; /* len bits being decoded */ |
133 | int first; /* first code of length len */ |
134 | int count; /* number of codes of length len */ |
135 | int index; /* index of first code of length len in symbol table */ |
136 | int bitbuf; /* bits from stream */ |
137 | int left; /* bits left in next or left to process */ |
138 | short *next; /* next number of codes */ |
139 | |
140 | bitbuf = s->bitbuf; |
141 | left = s->bitcnt; |
142 | code = first = index = 0; |
143 | len = 1; |
144 | next = h->count + 1; |
145 | while (1) { |
146 | while (left--) { |
147 | code |= (bitbuf & 1) ^ 1; /* invert code */ |
148 | bitbuf >>= 1; |
149 | count = *next++; |
150 | if (code < first + count) { /* if length len, return symbol */ |
151 | s->bitbuf = bitbuf; |
152 | s->bitcnt = (s->bitcnt - len) & 7; |
153 | return h->symbol[index + (code - first)]; |
154 | } |
155 | index += count; /* else update for next length */ |
156 | first += count; |
157 | first <<= 1; |
158 | code <<= 1; |
159 | len++; |
160 | } |
161 | left = (MAXBITS+1) - len; |
162 | if (left == 0) break; |
163 | if (s->left == 0) { |
164 | s->left = s->infun(s->inhow, &(s->in)); |
165 | if (s->left == 0) longjmp(s->env, 1); /* out of input */ |
166 | } |
167 | bitbuf = *(s->in)++; |
168 | s->left--; |
169 | if (left > 8) left = 8; |
170 | } |
171 | return -9; /* ran out of codes */ |
172 | } |
173 | |
174 | /* |
175 | * Given a list of repeated code lengths rep[0..n-1], where each byte is a |
176 | * count (high four bits + 1) and a code length (low four bits), generate the |
177 | * list of code lengths. This compaction reduces the size of the object code. |
178 | * Then given the list of code lengths length[0..n-1] representing a canonical |
179 | * Huffman code for n symbols, construct the tables required to decode those |
180 | * codes. Those tables are the number of codes of each length, and the symbols |
181 | * sorted by length, retaining their original order within each length. The |
182 | * return value is zero for a complete code set, negative for an over- |
183 | * subscribed code set, and positive for an incomplete code set. The tables |
184 | * can be used if the return value is zero or positive, but they cannot be used |
185 | * if the return value is negative. If the return value is zero, it is not |
186 | * possible for decode() using that table to return an error--any stream of |
187 | * enough bits will resolve to a symbol. If the return value is positive, then |
188 | * it is possible for decode() using that table to return an error for received |
189 | * codes past the end of the incomplete lengths. |
190 | */ |
191 | local int construct(struct huffman *h, const unsigned char *rep, int n) |
192 | { |
193 | int symbol; /* current symbol when stepping through length[] */ |
194 | int len; /* current length when stepping through h->count[] */ |
195 | int left; /* number of possible codes left of current length */ |
196 | short offs[MAXBITS+1]; /* offsets in symbol table for each length */ |
197 | short length[256]; /* code lengths */ |
198 | |
199 | /* convert compact repeat counts into symbol bit length list */ |
200 | symbol = 0; |
201 | do { |
202 | len = *rep++; |
203 | left = (len >> 4) + 1; |
204 | len &= 15; |
205 | do { |
206 | length[symbol++] = len; |
207 | } while (--left); |
208 | } while (--n); |
209 | n = symbol; |
210 | |
211 | /* count number of codes of each length */ |
212 | for (len = 0; len <= MAXBITS; len++) |
213 | h->count[len] = 0; |
214 | for (symbol = 0; symbol < n; symbol++) |
215 | (h->count[length[symbol]])++; /* assumes lengths are within bounds */ |
216 | if (h->count[0] == n) /* no codes! */ |
217 | return 0; /* complete, but decode() will fail */ |
218 | |
219 | /* check for an over-subscribed or incomplete set of lengths */ |
220 | left = 1; /* one possible code of zero length */ |
221 | for (len = 1; len <= MAXBITS; len++) { |
222 | left <<= 1; /* one more bit, double codes left */ |
223 | left -= h->count[len]; /* deduct count from possible codes */ |
224 | if (left < 0) return left; /* over-subscribed--return negative */ |
225 | } /* left > 0 means incomplete */ |
226 | |
227 | /* generate offsets into symbol table for each length for sorting */ |
228 | offs[1] = 0; |
229 | for (len = 1; len < MAXBITS; len++) |
230 | offs[len + 1] = offs[len] + h->count[len]; |
231 | |
232 | /* |
233 | * put symbols in table sorted by length, by symbol order within each |
234 | * length |
235 | */ |
236 | for (symbol = 0; symbol < n; symbol++) |
237 | if (length[symbol] != 0) |
238 | h->symbol[offs[length[symbol]]++] = symbol; |
239 | |
240 | /* return zero for complete set, positive for incomplete set */ |
241 | return left; |
242 | } |
243 | |
244 | /* |
245 | * Decode PKWare Compression Library stream. |
246 | * |
247 | * Format notes: |
248 | * |
249 | * - First byte is 0 if literals are uncoded or 1 if they are coded. Second |
250 | * byte is 4, 5, or 6 for the number of extra bits in the distance code. |
251 | * This is the base-2 logarithm of the dictionary size minus six. |
252 | * |
253 | * - Compressed data is a combination of literals and length/distance pairs |
254 | * terminated by an end code. Literals are either Huffman coded or |
255 | * uncoded bytes. A length/distance pair is a coded length followed by a |
256 | * coded distance to represent a string that occurs earlier in the |
257 | * uncompressed data that occurs again at the current location. |
258 | * |
259 | * - A bit preceding a literal or length/distance pair indicates which comes |
260 | * next, 0 for literals, 1 for length/distance. |
261 | * |
262 | * - If literals are uncoded, then the next eight bits are the literal, in the |
263 | * normal bit order in the stream, i.e. no bit-reversal is needed. Similarly, |
264 | * no bit reversal is needed for either the length extra bits or the distance |
265 | * extra bits. |
266 | * |
267 | * - Literal bytes are simply written to the output. A length/distance pair is |
268 | * an instruction to copy previously uncompressed bytes to the output. The |
269 | * copy is from distance bytes back in the output stream, copying for length |
270 | * bytes. |
271 | * |
272 | * - Distances pointing before the beginning of the output data are not |
273 | * permitted. |
274 | * |
275 | * - Overlapped copies, where the length is greater than the distance, are |
276 | * allowed and common. For example, a distance of one and a length of 518 |
277 | * simply copies the last byte 518 times. A distance of four and a length of |
278 | * twelve copies the last four bytes three times. A simple forward copy |
279 | * ignoring whether the length is greater than the distance or not implements |
280 | * this correctly. |
281 | */ |
282 | local int decomp(struct state *s) |
283 | { |
284 | int lit; /* true if literals are coded */ |
285 | int dict; /* log2(dictionary size) - 6 */ |
286 | int symbol; /* decoded symbol, extra bits for distance */ |
287 | int len; /* length for copy */ |
288 | unsigned dist; /* distance for copy */ |
289 | int copy; /* copy counter */ |
290 | unsigned char *from, *to; /* copy pointers */ |
291 | static int virgin = 1; /* build tables once */ |
292 | static short litcnt[MAXBITS+1], litsym[256]; /* litcode memory */ |
293 | static short lencnt[MAXBITS+1], lensym[16]; /* lencode memory */ |
294 | static short distcnt[MAXBITS+1], distsym[64]; /* distcode memory */ |
295 | static struct huffman litcode = {litcnt, litsym}; /* length code */ |
296 | static struct huffman lencode = {lencnt, lensym}; /* length code */ |
297 | static struct huffman distcode = {distcnt, distsym};/* distance code */ |
298 | /* bit lengths of literal codes */ |
299 | static const unsigned char litlen[] = { |
300 | 11, 124, 8, 7, 28, 7, 188, 13, 76, 4, 10, 8, 12, 10, 12, 10, 8, 23, 8, |
301 | 9, 7, 6, 7, 8, 7, 6, 55, 8, 23, 24, 12, 11, 7, 9, 11, 12, 6, 7, 22, 5, |
302 | 7, 24, 6, 11, 9, 6, 7, 22, 7, 11, 38, 7, 9, 8, 25, 11, 8, 11, 9, 12, |
303 | 8, 12, 5, 38, 5, 38, 5, 11, 7, 5, 6, 21, 6, 10, 53, 8, 7, 24, 10, 27, |
304 | 44, 253, 253, 253, 252, 252, 252, 13, 12, 45, 12, 45, 12, 61, 12, 45, |
305 | 44, 173}; |
306 | /* bit lengths of length codes 0..15 */ |
307 | static const unsigned char lenlen[] = {2, 35, 36, 53, 38, 23}; |
308 | /* bit lengths of distance codes 0..63 */ |
309 | static const unsigned char distlen[] = {2, 20, 53, 230, 247, 151, 248}; |
310 | static const short base[16] = { /* base for length codes */ |
311 | 3, 2, 4, 5, 6, 7, 8, 9, 10, 12, 16, 24, 40, 72, 136, 264}; |
312 | static const char extra[16] = { /* extra bits for length codes */ |
313 | 0, 0, 0, 0, 0, 0, 0, 0, 1, 2, 3, 4, 5, 6, 7, 8}; |
314 | |
315 | /* set up decoding tables (once--might not be thread-safe) */ |
316 | if (virgin) { |
317 | construct(&litcode, litlen, sizeof(litlen)); |
318 | construct(&lencode, lenlen, sizeof(lenlen)); |
319 | construct(&distcode, distlen, sizeof(distlen)); |
320 | virgin = 0; |
321 | } |
322 | |
323 | /* read header */ |
324 | lit = bits(s, 8); |
325 | if (lit > 1) return -1; |
326 | dict = bits(s, 8); |
327 | if (dict < 4 || dict > 6) return -2; |
328 | |
329 | /* decode literals and length/distance pairs */ |
330 | do { |
331 | if (bits(s, 1)) { |
332 | /* get length */ |
333 | symbol = decode(s, &lencode); |
334 | len = base[symbol] + bits(s, extra[symbol]); |
335 | if (len == 519) break; /* end code */ |
336 | |
337 | /* get distance */ |
338 | symbol = len == 2 ? 2 : dict; |
339 | dist = decode(s, &distcode) << symbol; |
340 | dist += bits(s, symbol); |
341 | dist++; |
342 | if (s->first && dist > s->next) |
343 | return -3; /* distance too far back */ |
344 | |
345 | /* copy length bytes from distance bytes back */ |
346 | do { |
347 | to = s->out + s->next; |
348 | from = to - dist; |
349 | copy = MAXWIN; |
350 | if (s->next < dist) { |
351 | from += copy; |
352 | copy = dist; |
353 | } |
354 | copy -= s->next; |
355 | if (copy > len) copy = len; |
356 | len -= copy; |
357 | s->next += copy; |
358 | do { |
359 | *to++ = *from++; |
360 | } while (--copy); |
361 | if (s->next == MAXWIN) { |
362 | if (s->outfun(s->outhow, s->out, s->next)) return 1; |
363 | s->next = 0; |
364 | s->first = 0; |
365 | } |
366 | } while (len != 0); |
367 | } |
368 | else { |
369 | /* get literal and write it */ |
370 | symbol = lit ? decode(s, &litcode) : bits(s, 8); |
371 | s->out[s->next++] = symbol; |
372 | if (s->next == MAXWIN) { |
373 | if (s->outfun(s->outhow, s->out, s->next)) return 1; |
374 | s->next = 0; |
375 | s->first = 0; |
376 | } |
377 | } |
378 | } while (1); |
379 | return 0; |
380 | } |
381 | |
382 | /* See comments in blast.h */ |
383 | int blast(blast_in infun, void *inhow, blast_out outfun, void *outhow, |
384 | unsigned *left, unsigned char **in) |
385 | { |
386 | struct state s; /* input/output state */ |
387 | int err; /* return value */ |
388 | |
389 | /* initialize input state */ |
390 | s.infun = infun; |
391 | s.inhow = inhow; |
392 | if (left != NULL && *left) { |
393 | s.left = *left; |
394 | s.in = *in; |
395 | } |
396 | else |
397 | s.left = 0; |
398 | s.bitbuf = 0; |
399 | s.bitcnt = 0; |
400 | |
401 | /* initialize output state */ |
402 | s.outfun = outfun; |
403 | s.outhow = outhow; |
404 | s.next = 0; |
405 | s.first = 1; |
406 | |
407 | /* return if bits() or decode() tries to read past available input */ |
408 | if (setjmp(s.env) != 0) /* if came back here via longjmp(), */ |
409 | err = 2; /* then skip decomp(), return error */ |
410 | else |
411 | err = decomp(&s); /* decompress */ |
412 | |
413 | /* return unused input */ |
414 | if (left != NULL) |
415 | *left = s.left; |
416 | if (in != NULL) |
417 | *in = s.left ? s.in : NULL; |
418 | |
419 | /* write any leftover output and update the error code if needed */ |
420 | if (err != 1 && s.next && s.outfun(s.outhow, s.out, s.next) && err == 0) |
421 | err = 1; |
422 | return err; |
423 | } |
424 | |
425 | #ifdef TEST |
426 | /* Example of how to use blast() */ |
427 | #include <stdio.h> |
428 | #include <stdlib.h> |
429 | |
430 | #define CHUNK 16384 |
431 | |
432 | local unsigned inf(void *how, unsigned char **buf) |
433 | { |
434 | static unsigned char hold[CHUNK]; |
435 | |
436 | *buf = hold; |
437 | return fread(hold, 1, CHUNK, (FILE *)how); |
438 | } |
439 | |
440 | local int outf(void *how, unsigned char *buf, unsigned len) |
441 | { |
442 | return fwrite(buf, 1, len, (FILE *)how) != len; |
443 | } |
444 | |
445 | /* Decompress a PKWare Compression Library stream from stdin to stdout */ |
446 | int main(void) |
447 | { |
448 | int ret; |
449 | unsigned left; |
450 | |
451 | /* decompress to stdout */ |
452 | left = 0; |
453 | ret = blast(inf, stdin, outf, stdout, &left, NULL); |
454 | if (ret != 0) |
455 | fprintf(stderr, "blast error: %d\n", ret); |
456 | |
457 | /* count any leftover bytes */ |
458 | while (getchar() != EOF) |
459 | left++; |
460 | if (left) |
461 | fprintf(stderr, "blast warning: %u unused bytes of input\n", left); |
462 | |
463 | /* return blast() error code */ |
464 | return ret; |
465 | } |
466 | #endif |