1 | /* vi: set sw=4 ts=4: */ |
---|
2 | /* |
---|
3 | * Gzip implementation for busybox |
---|
4 | * |
---|
5 | * Based on GNU gzip Copyright (C) 1992-1993 Jean-loup Gailly. |
---|
6 | * |
---|
7 | * Originally adjusted for busybox by Charles P. Wright <cpw@unix.asb.com> |
---|
8 | * "this is a stripped down version of gzip I put into busybox, it does |
---|
9 | * only standard in to standard out with -9 compression. It also requires |
---|
10 | * the zcat module for some important functions." |
---|
11 | * |
---|
12 | * Adjusted further by Erik Andersen <andersen@codepoet.org> to support |
---|
13 | * files as well as stdin/stdout, and to generally behave itself wrt |
---|
14 | * command line handling. |
---|
15 | * |
---|
16 | * Licensed under GPLv2 or later, see file LICENSE in this tarball for details. |
---|
17 | */ |
---|
18 | |
---|
19 | /* big objects in bss: |
---|
20 | * 00000020 b bl_count |
---|
21 | * 00000074 b base_length |
---|
22 | * 00000078 b base_dist |
---|
23 | * 00000078 b static_dtree |
---|
24 | * 0000009c b bl_tree |
---|
25 | * 000000f4 b dyn_dtree |
---|
26 | * 00000100 b length_code |
---|
27 | * 00000200 b dist_code |
---|
28 | * 0000023d b depth |
---|
29 | * 00000400 b flag_buf |
---|
30 | * 0000047a b heap |
---|
31 | * 00000480 b static_ltree |
---|
32 | * 000008f4 b dyn_ltree |
---|
33 | */ |
---|
34 | |
---|
35 | /* TODO: full support for -v for DESKTOP |
---|
36 | * "/usr/bin/gzip -v a bogus aa" should say: |
---|
37 | a: 85.1% -- replaced with a.gz |
---|
38 | gzip: bogus: No such file or directory |
---|
39 | aa: 85.1% -- replaced with aa.gz |
---|
40 | */ |
---|
41 | |
---|
42 | #include "libbb.h" |
---|
43 | |
---|
44 | |
---|
45 | /* =========================================================================== |
---|
46 | */ |
---|
47 | //#define DEBUG 1 |
---|
48 | /* Diagnostic functions */ |
---|
49 | #ifdef DEBUG |
---|
50 | # define Assert(cond,msg) { if (!(cond)) bb_error_msg(msg); } |
---|
51 | # define Trace(x) fprintf x |
---|
52 | # define Tracev(x) {if (verbose) fprintf x; } |
---|
53 | # define Tracevv(x) {if (verbose > 1) fprintf x; } |
---|
54 | # define Tracec(c,x) {if (verbose && (c)) fprintf x; } |
---|
55 | # define Tracecv(c,x) {if (verbose > 1 && (c)) fprintf x; } |
---|
56 | #else |
---|
57 | # define Assert(cond,msg) |
---|
58 | # define Trace(x) |
---|
59 | # define Tracev(x) |
---|
60 | # define Tracevv(x) |
---|
61 | # define Tracec(c,x) |
---|
62 | # define Tracecv(c,x) |
---|
63 | #endif |
---|
64 | |
---|
65 | |
---|
66 | /* =========================================================================== |
---|
67 | */ |
---|
68 | #define SMALL_MEM |
---|
69 | |
---|
70 | #ifndef INBUFSIZ |
---|
71 | # ifdef SMALL_MEM |
---|
72 | # define INBUFSIZ 0x2000 /* input buffer size */ |
---|
73 | # else |
---|
74 | # define INBUFSIZ 0x8000 /* input buffer size */ |
---|
75 | # endif |
---|
76 | #endif |
---|
77 | |
---|
78 | #ifndef OUTBUFSIZ |
---|
79 | # ifdef SMALL_MEM |
---|
80 | # define OUTBUFSIZ 8192 /* output buffer size */ |
---|
81 | # else |
---|
82 | # define OUTBUFSIZ 16384 /* output buffer size */ |
---|
83 | # endif |
---|
84 | #endif |
---|
85 | |
---|
86 | #ifndef DIST_BUFSIZE |
---|
87 | # ifdef SMALL_MEM |
---|
88 | # define DIST_BUFSIZE 0x2000 /* buffer for distances, see trees.c */ |
---|
89 | # else |
---|
90 | # define DIST_BUFSIZE 0x8000 /* buffer for distances, see trees.c */ |
---|
91 | # endif |
---|
92 | #endif |
---|
93 | |
---|
94 | /* gzip flag byte */ |
---|
95 | #define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */ |
---|
96 | #define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */ |
---|
97 | #define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */ |
---|
98 | #define ORIG_NAME 0x08 /* bit 3 set: original file name present */ |
---|
99 | #define COMMENT 0x10 /* bit 4 set: file comment present */ |
---|
100 | #define RESERVED 0xC0 /* bit 6,7: reserved */ |
---|
101 | |
---|
102 | /* internal file attribute */ |
---|
103 | #define UNKNOWN 0xffff |
---|
104 | #define BINARY 0 |
---|
105 | #define ASCII 1 |
---|
106 | |
---|
107 | #ifndef WSIZE |
---|
108 | # define WSIZE 0x8000 /* window size--must be a power of two, and */ |
---|
109 | #endif /* at least 32K for zip's deflate method */ |
---|
110 | |
---|
111 | #define MIN_MATCH 3 |
---|
112 | #define MAX_MATCH 258 |
---|
113 | /* The minimum and maximum match lengths */ |
---|
114 | |
---|
115 | #define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1) |
---|
116 | /* Minimum amount of lookahead, except at the end of the input file. |
---|
117 | * See deflate.c for comments about the MIN_MATCH+1. |
---|
118 | */ |
---|
119 | |
---|
120 | #define MAX_DIST (WSIZE-MIN_LOOKAHEAD) |
---|
121 | /* In order to simplify the code, particularly on 16 bit machines, match |
---|
122 | * distances are limited to MAX_DIST instead of WSIZE. |
---|
123 | */ |
---|
124 | |
---|
125 | #ifndef MAX_PATH_LEN |
---|
126 | # define MAX_PATH_LEN 1024 /* max pathname length */ |
---|
127 | #endif |
---|
128 | |
---|
129 | #define seekable() 0 /* force sequential output */ |
---|
130 | #define translate_eol 0 /* no option -a yet */ |
---|
131 | |
---|
132 | #ifndef BITS |
---|
133 | # define BITS 16 |
---|
134 | #endif |
---|
135 | #define INIT_BITS 9 /* Initial number of bits per code */ |
---|
136 | |
---|
137 | #define BIT_MASK 0x1f /* Mask for 'number of compression bits' */ |
---|
138 | /* Mask 0x20 is reserved to mean a fourth header byte, and 0x40 is free. |
---|
139 | * It's a pity that old uncompress does not check bit 0x20. That makes |
---|
140 | * extension of the format actually undesirable because old compress |
---|
141 | * would just crash on the new format instead of giving a meaningful |
---|
142 | * error message. It does check the number of bits, but it's more |
---|
143 | * helpful to say "unsupported format, get a new version" than |
---|
144 | * "can only handle 16 bits". |
---|
145 | */ |
---|
146 | |
---|
147 | #ifdef MAX_EXT_CHARS |
---|
148 | # define MAX_SUFFIX MAX_EXT_CHARS |
---|
149 | #else |
---|
150 | # define MAX_SUFFIX 30 |
---|
151 | #endif |
---|
152 | |
---|
153 | |
---|
154 | /* =========================================================================== |
---|
155 | * Compile with MEDIUM_MEM to reduce the memory requirements or |
---|
156 | * with SMALL_MEM to use as little memory as possible. Use BIG_MEM if the |
---|
157 | * entire input file can be held in memory (not possible on 16 bit systems). |
---|
158 | * Warning: defining these symbols affects HASH_BITS (see below) and thus |
---|
159 | * affects the compression ratio. The compressed output |
---|
160 | * is still correct, and might even be smaller in some cases. |
---|
161 | */ |
---|
162 | |
---|
163 | #ifdef SMALL_MEM |
---|
164 | # define HASH_BITS 13 /* Number of bits used to hash strings */ |
---|
165 | #endif |
---|
166 | #ifdef MEDIUM_MEM |
---|
167 | # define HASH_BITS 14 |
---|
168 | #endif |
---|
169 | #ifndef HASH_BITS |
---|
170 | # define HASH_BITS 15 |
---|
171 | /* For portability to 16 bit machines, do not use values above 15. */ |
---|
172 | #endif |
---|
173 | |
---|
174 | #define HASH_SIZE (unsigned)(1<<HASH_BITS) |
---|
175 | #define HASH_MASK (HASH_SIZE-1) |
---|
176 | #define WMASK (WSIZE-1) |
---|
177 | /* HASH_SIZE and WSIZE must be powers of two */ |
---|
178 | #ifndef TOO_FAR |
---|
179 | # define TOO_FAR 4096 |
---|
180 | #endif |
---|
181 | /* Matches of length 3 are discarded if their distance exceeds TOO_FAR */ |
---|
182 | |
---|
183 | |
---|
184 | /* =========================================================================== |
---|
185 | * These types are not really 'char', 'short' and 'long' |
---|
186 | */ |
---|
187 | typedef uint8_t uch; |
---|
188 | typedef uint16_t ush; |
---|
189 | typedef uint32_t ulg; |
---|
190 | typedef int32_t lng; |
---|
191 | |
---|
192 | typedef ush Pos; |
---|
193 | typedef unsigned IPos; |
---|
194 | /* A Pos is an index in the character window. We use short instead of int to |
---|
195 | * save space in the various tables. IPos is used only for parameter passing. |
---|
196 | */ |
---|
197 | |
---|
198 | enum { |
---|
199 | WINDOW_SIZE = 2 * WSIZE, |
---|
200 | /* window size, 2*WSIZE except for MMAP or BIG_MEM, where it is the |
---|
201 | * input file length plus MIN_LOOKAHEAD. |
---|
202 | */ |
---|
203 | |
---|
204 | max_chain_length = 4096, |
---|
205 | /* To speed up deflation, hash chains are never searched beyond this length. |
---|
206 | * A higher limit improves compression ratio but degrades the speed. |
---|
207 | */ |
---|
208 | |
---|
209 | max_lazy_match = 258, |
---|
210 | /* Attempt to find a better match only when the current match is strictly |
---|
211 | * smaller than this value. This mechanism is used only for compression |
---|
212 | * levels >= 4. |
---|
213 | */ |
---|
214 | |
---|
215 | max_insert_length = max_lazy_match, |
---|
216 | /* Insert new strings in the hash table only if the match length |
---|
217 | * is not greater than this length. This saves time but degrades compression. |
---|
218 | * max_insert_length is used only for compression levels <= 3. |
---|
219 | */ |
---|
220 | |
---|
221 | good_match = 32, |
---|
222 | /* Use a faster search when the previous match is longer than this */ |
---|
223 | |
---|
224 | /* Values for max_lazy_match, good_match and max_chain_length, depending on |
---|
225 | * the desired pack level (0..9). The values given below have been tuned to |
---|
226 | * exclude worst case performance for pathological files. Better values may be |
---|
227 | * found for specific files. |
---|
228 | */ |
---|
229 | |
---|
230 | nice_match = 258, /* Stop searching when current match exceeds this */ |
---|
231 | /* Note: the deflate() code requires max_lazy >= MIN_MATCH and max_chain >= 4 |
---|
232 | * For deflate_fast() (levels <= 3) good is ignored and lazy has a different |
---|
233 | * meaning. |
---|
234 | */ |
---|
235 | }; |
---|
236 | |
---|
237 | |
---|
238 | struct globals { |
---|
239 | |
---|
240 | lng block_start; |
---|
241 | |
---|
242 | /* window position at the beginning of the current output block. Gets |
---|
243 | * negative when the window is moved backwards. |
---|
244 | */ |
---|
245 | unsigned ins_h; /* hash index of string to be inserted */ |
---|
246 | |
---|
247 | #define H_SHIFT ((HASH_BITS+MIN_MATCH-1) / MIN_MATCH) |
---|
248 | /* Number of bits by which ins_h and del_h must be shifted at each |
---|
249 | * input step. It must be such that after MIN_MATCH steps, the oldest |
---|
250 | * byte no longer takes part in the hash key, that is: |
---|
251 | * H_SHIFT * MIN_MATCH >= HASH_BITS |
---|
252 | */ |
---|
253 | |
---|
254 | unsigned prev_length; |
---|
255 | |
---|
256 | /* Length of the best match at previous step. Matches not greater than this |
---|
257 | * are discarded. This is used in the lazy match evaluation. |
---|
258 | */ |
---|
259 | |
---|
260 | unsigned strstart; /* start of string to insert */ |
---|
261 | unsigned match_start; /* start of matching string */ |
---|
262 | unsigned lookahead; /* number of valid bytes ahead in window */ |
---|
263 | |
---|
264 | /* =========================================================================== |
---|
265 | */ |
---|
266 | #define DECLARE(type, array, size) \ |
---|
267 | type * array |
---|
268 | #define ALLOC(type, array, size) \ |
---|
269 | array = xzalloc((size_t)(((size)+1L)/2) * 2*sizeof(type)); |
---|
270 | #define FREE(array) \ |
---|
271 | do { free(array); array = NULL; } while (0) |
---|
272 | |
---|
273 | /* global buffers */ |
---|
274 | |
---|
275 | /* buffer for literals or lengths */ |
---|
276 | /* DECLARE(uch, l_buf, LIT_BUFSIZE); */ |
---|
277 | DECLARE(uch, l_buf, INBUFSIZ); |
---|
278 | |
---|
279 | DECLARE(ush, d_buf, DIST_BUFSIZE); |
---|
280 | DECLARE(uch, outbuf, OUTBUFSIZ); |
---|
281 | |
---|
282 | /* Sliding window. Input bytes are read into the second half of the window, |
---|
283 | * and move to the first half later to keep a dictionary of at least WSIZE |
---|
284 | * bytes. With this organization, matches are limited to a distance of |
---|
285 | * WSIZE-MAX_MATCH bytes, but this ensures that IO is always |
---|
286 | * performed with a length multiple of the block size. Also, it limits |
---|
287 | * the window size to 64K, which is quite useful on MSDOS. |
---|
288 | * To do: limit the window size to WSIZE+BSZ if SMALL_MEM (the code would |
---|
289 | * be less efficient). |
---|
290 | */ |
---|
291 | DECLARE(uch, window, 2L * WSIZE); |
---|
292 | |
---|
293 | /* Link to older string with same hash index. To limit the size of this |
---|
294 | * array to 64K, this link is maintained only for the last 32K strings. |
---|
295 | * An index in this array is thus a window index modulo 32K. |
---|
296 | */ |
---|
297 | /* DECLARE(Pos, prev, WSIZE); */ |
---|
298 | DECLARE(ush, prev, 1L << BITS); |
---|
299 | |
---|
300 | /* Heads of the hash chains or 0. */ |
---|
301 | /* DECLARE(Pos, head, 1<<HASH_BITS); */ |
---|
302 | #define head (G1.prev + WSIZE) /* hash head (see deflate.c) */ |
---|
303 | |
---|
304 | /* number of input bytes */ |
---|
305 | ulg isize; /* only 32 bits stored in .gz file */ |
---|
306 | |
---|
307 | /* bbox always use stdin/stdout */ |
---|
308 | #define ifd STDIN_FILENO /* input file descriptor */ |
---|
309 | #define ofd STDOUT_FILENO /* output file descriptor */ |
---|
310 | |
---|
311 | #ifdef DEBUG |
---|
312 | unsigned insize; /* valid bytes in l_buf */ |
---|
313 | #endif |
---|
314 | unsigned outcnt; /* bytes in output buffer */ |
---|
315 | |
---|
316 | smallint eofile; /* flag set at end of input file */ |
---|
317 | |
---|
318 | /* =========================================================================== |
---|
319 | * Local data used by the "bit string" routines. |
---|
320 | */ |
---|
321 | |
---|
322 | unsigned short bi_buf; |
---|
323 | |
---|
324 | /* Output buffer. bits are inserted starting at the bottom (least significant |
---|
325 | * bits). |
---|
326 | */ |
---|
327 | |
---|
328 | #undef BUF_SIZE |
---|
329 | #define BUF_SIZE (8 * sizeof(G1.bi_buf)) |
---|
330 | /* Number of bits used within bi_buf. (bi_buf might be implemented on |
---|
331 | * more than 16 bits on some systems.) |
---|
332 | */ |
---|
333 | |
---|
334 | int bi_valid; |
---|
335 | |
---|
336 | /* Current input function. Set to mem_read for in-memory compression */ |
---|
337 | |
---|
338 | #ifdef DEBUG |
---|
339 | ulg bits_sent; /* bit length of the compressed data */ |
---|
340 | #endif |
---|
341 | |
---|
342 | uint32_t *crc_32_tab; |
---|
343 | uint32_t crc; /* shift register contents */ |
---|
344 | }; |
---|
345 | |
---|
346 | #define G1 (*(ptr_to_globals - 1)) |
---|
347 | |
---|
348 | |
---|
349 | /* =========================================================================== |
---|
350 | * Write the output buffer outbuf[0..outcnt-1] and update bytes_out. |
---|
351 | * (used for the compressed data only) |
---|
352 | */ |
---|
353 | static void flush_outbuf(void) |
---|
354 | { |
---|
355 | if (G1.outcnt == 0) |
---|
356 | return; |
---|
357 | |
---|
358 | xwrite(ofd, (char *) G1.outbuf, G1.outcnt); |
---|
359 | G1.outcnt = 0; |
---|
360 | } |
---|
361 | |
---|
362 | |
---|
363 | /* =========================================================================== |
---|
364 | */ |
---|
365 | /* put_8bit is used for the compressed output */ |
---|
366 | #define put_8bit(c) \ |
---|
367 | do { \ |
---|
368 | G1.outbuf[G1.outcnt++] = (c); \ |
---|
369 | if (G1.outcnt == OUTBUFSIZ) flush_outbuf(); \ |
---|
370 | } while (0) |
---|
371 | |
---|
372 | /* Output a 16 bit value, lsb first */ |
---|
373 | static void put_16bit(ush w) |
---|
374 | { |
---|
375 | if (G1.outcnt < OUTBUFSIZ - 2) { |
---|
376 | G1.outbuf[G1.outcnt++] = w; |
---|
377 | G1.outbuf[G1.outcnt++] = w >> 8; |
---|
378 | } else { |
---|
379 | put_8bit(w); |
---|
380 | put_8bit(w >> 8); |
---|
381 | } |
---|
382 | } |
---|
383 | |
---|
384 | static void put_32bit(ulg n) |
---|
385 | { |
---|
386 | put_16bit(n); |
---|
387 | put_16bit(n >> 16); |
---|
388 | } |
---|
389 | |
---|
390 | /* =========================================================================== |
---|
391 | * Clear input and output buffers |
---|
392 | */ |
---|
393 | static void clear_bufs(void) |
---|
394 | { |
---|
395 | G1.outcnt = 0; |
---|
396 | #ifdef DEBUG |
---|
397 | G1.insize = 0; |
---|
398 | #endif |
---|
399 | G1.isize = 0; |
---|
400 | } |
---|
401 | |
---|
402 | |
---|
403 | /* =========================================================================== |
---|
404 | * Run a set of bytes through the crc shift register. If s is a NULL |
---|
405 | * pointer, then initialize the crc shift register contents instead. |
---|
406 | * Return the current crc in either case. |
---|
407 | */ |
---|
408 | static uint32_t updcrc(uch * s, unsigned n) |
---|
409 | { |
---|
410 | uint32_t c = G1.crc; |
---|
411 | while (n) { |
---|
412 | c = G1.crc_32_tab[(uch)(c ^ *s++)] ^ (c >> 8); |
---|
413 | n--; |
---|
414 | } |
---|
415 | G1.crc = c; |
---|
416 | return c; |
---|
417 | } |
---|
418 | |
---|
419 | |
---|
420 | /* =========================================================================== |
---|
421 | * Read a new buffer from the current input file, perform end-of-line |
---|
422 | * translation, and update the crc and input file size. |
---|
423 | * IN assertion: size >= 2 (for end-of-line translation) |
---|
424 | */ |
---|
425 | static unsigned file_read(void *buf, unsigned size) |
---|
426 | { |
---|
427 | unsigned len; |
---|
428 | |
---|
429 | Assert(G1.insize == 0, "l_buf not empty"); |
---|
430 | |
---|
431 | len = safe_read(ifd, buf, size); |
---|
432 | if (len == (unsigned)(-1) || len == 0) |
---|
433 | return len; |
---|
434 | |
---|
435 | updcrc(buf, len); |
---|
436 | G1.isize += len; |
---|
437 | return len; |
---|
438 | } |
---|
439 | |
---|
440 | |
---|
441 | /* =========================================================================== |
---|
442 | * Send a value on a given number of bits. |
---|
443 | * IN assertion: length <= 16 and value fits in length bits. |
---|
444 | */ |
---|
445 | static void send_bits(int value, int length) |
---|
446 | { |
---|
447 | #ifdef DEBUG |
---|
448 | Tracev((stderr, " l %2d v %4x ", length, value)); |
---|
449 | Assert(length > 0 && length <= 15, "invalid length"); |
---|
450 | G1.bits_sent += length; |
---|
451 | #endif |
---|
452 | /* If not enough room in bi_buf, use (valid) bits from bi_buf and |
---|
453 | * (16 - bi_valid) bits from value, leaving (width - (16-bi_valid)) |
---|
454 | * unused bits in value. |
---|
455 | */ |
---|
456 | if (G1.bi_valid > (int) BUF_SIZE - length) { |
---|
457 | G1.bi_buf |= (value << G1.bi_valid); |
---|
458 | put_16bit(G1.bi_buf); |
---|
459 | G1.bi_buf = (ush) value >> (BUF_SIZE - G1.bi_valid); |
---|
460 | G1.bi_valid += length - BUF_SIZE; |
---|
461 | } else { |
---|
462 | G1.bi_buf |= value << G1.bi_valid; |
---|
463 | G1.bi_valid += length; |
---|
464 | } |
---|
465 | } |
---|
466 | |
---|
467 | |
---|
468 | /* =========================================================================== |
---|
469 | * Reverse the first len bits of a code, using straightforward code (a faster |
---|
470 | * method would use a table) |
---|
471 | * IN assertion: 1 <= len <= 15 |
---|
472 | */ |
---|
473 | static unsigned bi_reverse(unsigned code, int len) |
---|
474 | { |
---|
475 | unsigned res = 0; |
---|
476 | |
---|
477 | while (1) { |
---|
478 | res |= code & 1; |
---|
479 | if (--len <= 0) return res; |
---|
480 | code >>= 1; |
---|
481 | res <<= 1; |
---|
482 | } |
---|
483 | } |
---|
484 | |
---|
485 | |
---|
486 | /* =========================================================================== |
---|
487 | * Write out any remaining bits in an incomplete byte. |
---|
488 | */ |
---|
489 | static void bi_windup(void) |
---|
490 | { |
---|
491 | if (G1.bi_valid > 8) { |
---|
492 | put_16bit(G1.bi_buf); |
---|
493 | } else if (G1.bi_valid > 0) { |
---|
494 | put_8bit(G1.bi_buf); |
---|
495 | } |
---|
496 | G1.bi_buf = 0; |
---|
497 | G1.bi_valid = 0; |
---|
498 | #ifdef DEBUG |
---|
499 | G1.bits_sent = (G1.bits_sent + 7) & ~7; |
---|
500 | #endif |
---|
501 | } |
---|
502 | |
---|
503 | |
---|
504 | /* =========================================================================== |
---|
505 | * Copy a stored block to the zip file, storing first the length and its |
---|
506 | * one's complement if requested. |
---|
507 | */ |
---|
508 | static void copy_block(char *buf, unsigned len, int header) |
---|
509 | { |
---|
510 | bi_windup(); /* align on byte boundary */ |
---|
511 | |
---|
512 | if (header) { |
---|
513 | put_16bit(len); |
---|
514 | put_16bit(~len); |
---|
515 | #ifdef DEBUG |
---|
516 | G1.bits_sent += 2 * 16; |
---|
517 | #endif |
---|
518 | } |
---|
519 | #ifdef DEBUG |
---|
520 | G1.bits_sent += (ulg) len << 3; |
---|
521 | #endif |
---|
522 | while (len--) { |
---|
523 | put_8bit(*buf++); |
---|
524 | } |
---|
525 | } |
---|
526 | |
---|
527 | |
---|
528 | /* =========================================================================== |
---|
529 | * Fill the window when the lookahead becomes insufficient. |
---|
530 | * Updates strstart and lookahead, and sets eofile if end of input file. |
---|
531 | * IN assertion: lookahead < MIN_LOOKAHEAD && strstart + lookahead > 0 |
---|
532 | * OUT assertions: at least one byte has been read, or eofile is set; |
---|
533 | * file reads are performed for at least two bytes (required for the |
---|
534 | * translate_eol option). |
---|
535 | */ |
---|
536 | static void fill_window(void) |
---|
537 | { |
---|
538 | unsigned n, m; |
---|
539 | unsigned more = WINDOW_SIZE - G1.lookahead - G1.strstart; |
---|
540 | /* Amount of free space at the end of the window. */ |
---|
541 | |
---|
542 | /* If the window is almost full and there is insufficient lookahead, |
---|
543 | * move the upper half to the lower one to make room in the upper half. |
---|
544 | */ |
---|
545 | if (more == (unsigned) -1) { |
---|
546 | /* Very unlikely, but possible on 16 bit machine if strstart == 0 |
---|
547 | * and lookahead == 1 (input done one byte at time) |
---|
548 | */ |
---|
549 | more--; |
---|
550 | } else if (G1.strstart >= WSIZE + MAX_DIST) { |
---|
551 | /* By the IN assertion, the window is not empty so we can't confuse |
---|
552 | * more == 0 with more == 64K on a 16 bit machine. |
---|
553 | */ |
---|
554 | Assert(WINDOW_SIZE == 2 * WSIZE, "no sliding with BIG_MEM"); |
---|
555 | |
---|
556 | memcpy(G1.window, G1.window + WSIZE, WSIZE); |
---|
557 | G1.match_start -= WSIZE; |
---|
558 | G1.strstart -= WSIZE; /* we now have strstart >= MAX_DIST: */ |
---|
559 | |
---|
560 | G1.block_start -= WSIZE; |
---|
561 | |
---|
562 | for (n = 0; n < HASH_SIZE; n++) { |
---|
563 | m = head[n]; |
---|
564 | head[n] = (Pos) (m >= WSIZE ? m - WSIZE : 0); |
---|
565 | } |
---|
566 | for (n = 0; n < WSIZE; n++) { |
---|
567 | m = G1.prev[n]; |
---|
568 | G1.prev[n] = (Pos) (m >= WSIZE ? m - WSIZE : 0); |
---|
569 | /* If n is not on any hash chain, prev[n] is garbage but |
---|
570 | * its value will never be used. |
---|
571 | */ |
---|
572 | } |
---|
573 | more += WSIZE; |
---|
574 | } |
---|
575 | /* At this point, more >= 2 */ |
---|
576 | if (!G1.eofile) { |
---|
577 | n = file_read(G1.window + G1.strstart + G1.lookahead, more); |
---|
578 | if (n == 0 || n == (unsigned) -1) { |
---|
579 | G1.eofile = 1; |
---|
580 | } else { |
---|
581 | G1.lookahead += n; |
---|
582 | } |
---|
583 | } |
---|
584 | } |
---|
585 | |
---|
586 | |
---|
587 | /* =========================================================================== |
---|
588 | * Set match_start to the longest match starting at the given string and |
---|
589 | * return its length. Matches shorter or equal to prev_length are discarded, |
---|
590 | * in which case the result is equal to prev_length and match_start is |
---|
591 | * garbage. |
---|
592 | * IN assertions: cur_match is the head of the hash chain for the current |
---|
593 | * string (strstart) and its distance is <= MAX_DIST, and prev_length >= 1 |
---|
594 | */ |
---|
595 | |
---|
596 | /* For MSDOS, OS/2 and 386 Unix, an optimized version is in match.asm or |
---|
597 | * match.s. The code is functionally equivalent, so you can use the C version |
---|
598 | * if desired. |
---|
599 | */ |
---|
600 | static int longest_match(IPos cur_match) |
---|
601 | { |
---|
602 | unsigned chain_length = max_chain_length; /* max hash chain length */ |
---|
603 | uch *scan = G1.window + G1.strstart; /* current string */ |
---|
604 | uch *match; /* matched string */ |
---|
605 | int len; /* length of current match */ |
---|
606 | int best_len = G1.prev_length; /* best match length so far */ |
---|
607 | IPos limit = G1.strstart > (IPos) MAX_DIST ? G1.strstart - (IPos) MAX_DIST : 0; |
---|
608 | /* Stop when cur_match becomes <= limit. To simplify the code, |
---|
609 | * we prevent matches with the string of window index 0. |
---|
610 | */ |
---|
611 | |
---|
612 | /* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16. |
---|
613 | * It is easy to get rid of this optimization if necessary. |
---|
614 | */ |
---|
615 | #if HASH_BITS < 8 || MAX_MATCH != 258 |
---|
616 | # error Code too clever |
---|
617 | #endif |
---|
618 | uch *strend = G1.window + G1.strstart + MAX_MATCH; |
---|
619 | uch scan_end1 = scan[best_len - 1]; |
---|
620 | uch scan_end = scan[best_len]; |
---|
621 | |
---|
622 | /* Do not waste too much time if we already have a good match: */ |
---|
623 | if (G1.prev_length >= good_match) { |
---|
624 | chain_length >>= 2; |
---|
625 | } |
---|
626 | Assert(G1.strstart <= WINDOW_SIZE - MIN_LOOKAHEAD, "insufficient lookahead"); |
---|
627 | |
---|
628 | do { |
---|
629 | Assert(cur_match < G1.strstart, "no future"); |
---|
630 | match = G1.window + cur_match; |
---|
631 | |
---|
632 | /* Skip to next match if the match length cannot increase |
---|
633 | * or if the match length is less than 2: |
---|
634 | */ |
---|
635 | if (match[best_len] != scan_end || |
---|
636 | match[best_len - 1] != scan_end1 || |
---|
637 | *match != *scan || *++match != scan[1]) |
---|
638 | continue; |
---|
639 | |
---|
640 | /* The check at best_len-1 can be removed because it will be made |
---|
641 | * again later. (This heuristic is not always a win.) |
---|
642 | * It is not necessary to compare scan[2] and match[2] since they |
---|
643 | * are always equal when the other bytes match, given that |
---|
644 | * the hash keys are equal and that HASH_BITS >= 8. |
---|
645 | */ |
---|
646 | scan += 2, match++; |
---|
647 | |
---|
648 | /* We check for insufficient lookahead only every 8th comparison; |
---|
649 | * the 256th check will be made at strstart+258. |
---|
650 | */ |
---|
651 | do { |
---|
652 | } while (*++scan == *++match && *++scan == *++match && |
---|
653 | *++scan == *++match && *++scan == *++match && |
---|
654 | *++scan == *++match && *++scan == *++match && |
---|
655 | *++scan == *++match && *++scan == *++match && scan < strend); |
---|
656 | |
---|
657 | len = MAX_MATCH - (int) (strend - scan); |
---|
658 | scan = strend - MAX_MATCH; |
---|
659 | |
---|
660 | if (len > best_len) { |
---|
661 | G1.match_start = cur_match; |
---|
662 | best_len = len; |
---|
663 | if (len >= nice_match) |
---|
664 | break; |
---|
665 | scan_end1 = scan[best_len - 1]; |
---|
666 | scan_end = scan[best_len]; |
---|
667 | } |
---|
668 | } while ((cur_match = G1.prev[cur_match & WMASK]) > limit |
---|
669 | && --chain_length != 0); |
---|
670 | |
---|
671 | return best_len; |
---|
672 | } |
---|
673 | |
---|
674 | |
---|
675 | #ifdef DEBUG |
---|
676 | /* =========================================================================== |
---|
677 | * Check that the match at match_start is indeed a match. |
---|
678 | */ |
---|
679 | static void check_match(IPos start, IPos match, int length) |
---|
680 | { |
---|
681 | /* check that the match is indeed a match */ |
---|
682 | if (memcmp(G1.window + match, G1.window + start, length) != 0) { |
---|
683 | bb_error_msg(" start %d, match %d, length %d", start, match, length); |
---|
684 | bb_error_msg("invalid match"); |
---|
685 | } |
---|
686 | if (verbose > 1) { |
---|
687 | bb_error_msg("\\[%d,%d]", start - match, length); |
---|
688 | do { |
---|
689 | putc(G1.window[start++], stderr); |
---|
690 | } while (--length != 0); |
---|
691 | } |
---|
692 | } |
---|
693 | #else |
---|
694 | # define check_match(start, match, length) ((void)0) |
---|
695 | #endif |
---|
696 | |
---|
697 | |
---|
698 | /* trees.c -- output deflated data using Huffman coding |
---|
699 | * Copyright (C) 1992-1993 Jean-loup Gailly |
---|
700 | * This is free software; you can redistribute it and/or modify it under the |
---|
701 | * terms of the GNU General Public License, see the file COPYING. |
---|
702 | */ |
---|
703 | |
---|
704 | /* PURPOSE |
---|
705 | * Encode various sets of source values using variable-length |
---|
706 | * binary code trees. |
---|
707 | * |
---|
708 | * DISCUSSION |
---|
709 | * The PKZIP "deflation" process uses several Huffman trees. The more |
---|
710 | * common source values are represented by shorter bit sequences. |
---|
711 | * |
---|
712 | * Each code tree is stored in the ZIP file in a compressed form |
---|
713 | * which is itself a Huffman encoding of the lengths of |
---|
714 | * all the code strings (in ascending order by source values). |
---|
715 | * The actual code strings are reconstructed from the lengths in |
---|
716 | * the UNZIP process, as described in the "application note" |
---|
717 | * (APPNOTE.TXT) distributed as part of PKWARE's PKZIP program. |
---|
718 | * |
---|
719 | * REFERENCES |
---|
720 | * Lynch, Thomas J. |
---|
721 | * Data Compression: Techniques and Applications, pp. 53-55. |
---|
722 | * Lifetime Learning Publications, 1985. ISBN 0-534-03418-7. |
---|
723 | * |
---|
724 | * Storer, James A. |
---|
725 | * Data Compression: Methods and Theory, pp. 49-50. |
---|
726 | * Computer Science Press, 1988. ISBN 0-7167-8156-5. |
---|
727 | * |
---|
728 | * Sedgewick, R. |
---|
729 | * Algorithms, p290. |
---|
730 | * Addison-Wesley, 1983. ISBN 0-201-06672-6. |
---|
731 | * |
---|
732 | * INTERFACE |
---|
733 | * void ct_init() |
---|
734 | * Allocate the match buffer, initialize the various tables [and save |
---|
735 | * the location of the internal file attribute (ascii/binary) and |
---|
736 | * method (DEFLATE/STORE) -- deleted in bbox] |
---|
737 | * |
---|
738 | * void ct_tally(int dist, int lc); |
---|
739 | * Save the match info and tally the frequency counts. |
---|
740 | * |
---|
741 | * ulg flush_block(char *buf, ulg stored_len, int eof) |
---|
742 | * Determine the best encoding for the current block: dynamic trees, |
---|
743 | * static trees or store, and output the encoded block to the zip |
---|
744 | * file. Returns the total compressed length for the file so far. |
---|
745 | */ |
---|
746 | |
---|
747 | #define MAX_BITS 15 |
---|
748 | /* All codes must not exceed MAX_BITS bits */ |
---|
749 | |
---|
750 | #define MAX_BL_BITS 7 |
---|
751 | /* Bit length codes must not exceed MAX_BL_BITS bits */ |
---|
752 | |
---|
753 | #define LENGTH_CODES 29 |
---|
754 | /* number of length codes, not counting the special END_BLOCK code */ |
---|
755 | |
---|
756 | #define LITERALS 256 |
---|
757 | /* number of literal bytes 0..255 */ |
---|
758 | |
---|
759 | #define END_BLOCK 256 |
---|
760 | /* end of block literal code */ |
---|
761 | |
---|
762 | #define L_CODES (LITERALS+1+LENGTH_CODES) |
---|
763 | /* number of Literal or Length codes, including the END_BLOCK code */ |
---|
764 | |
---|
765 | #define D_CODES 30 |
---|
766 | /* number of distance codes */ |
---|
767 | |
---|
768 | #define BL_CODES 19 |
---|
769 | /* number of codes used to transfer the bit lengths */ |
---|
770 | |
---|
771 | /* extra bits for each length code */ |
---|
772 | static const uint8_t extra_lbits[LENGTH_CODES] ALIGN1 = { |
---|
773 | 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, |
---|
774 | 4, 4, 5, 5, 5, 5, 0 |
---|
775 | }; |
---|
776 | |
---|
777 | /* extra bits for each distance code */ |
---|
778 | static const uint8_t extra_dbits[D_CODES] ALIGN1 = { |
---|
779 | 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9, |
---|
780 | 10, 10, 11, 11, 12, 12, 13, 13 |
---|
781 | }; |
---|
782 | |
---|
783 | /* extra bits for each bit length code */ |
---|
784 | static const uint8_t extra_blbits[BL_CODES] ALIGN1 = { |
---|
785 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 3, 7 }; |
---|
786 | |
---|
787 | /* number of codes at each bit length for an optimal tree */ |
---|
788 | static const uint8_t bl_order[BL_CODES] ALIGN1 = { |
---|
789 | 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15 }; |
---|
790 | |
---|
791 | #define STORED_BLOCK 0 |
---|
792 | #define STATIC_TREES 1 |
---|
793 | #define DYN_TREES 2 |
---|
794 | /* The three kinds of block type */ |
---|
795 | |
---|
796 | #ifndef LIT_BUFSIZE |
---|
797 | # ifdef SMALL_MEM |
---|
798 | # define LIT_BUFSIZE 0x2000 |
---|
799 | # else |
---|
800 | # ifdef MEDIUM_MEM |
---|
801 | # define LIT_BUFSIZE 0x4000 |
---|
802 | # else |
---|
803 | # define LIT_BUFSIZE 0x8000 |
---|
804 | # endif |
---|
805 | # endif |
---|
806 | #endif |
---|
807 | #ifndef DIST_BUFSIZE |
---|
808 | # define DIST_BUFSIZE LIT_BUFSIZE |
---|
809 | #endif |
---|
810 | /* Sizes of match buffers for literals/lengths and distances. There are |
---|
811 | * 4 reasons for limiting LIT_BUFSIZE to 64K: |
---|
812 | * - frequencies can be kept in 16 bit counters |
---|
813 | * - if compression is not successful for the first block, all input data is |
---|
814 | * still in the window so we can still emit a stored block even when input |
---|
815 | * comes from standard input. (This can also be done for all blocks if |
---|
816 | * LIT_BUFSIZE is not greater than 32K.) |
---|
817 | * - if compression is not successful for a file smaller than 64K, we can |
---|
818 | * even emit a stored file instead of a stored block (saving 5 bytes). |
---|
819 | * - creating new Huffman trees less frequently may not provide fast |
---|
820 | * adaptation to changes in the input data statistics. (Take for |
---|
821 | * example a binary file with poorly compressible code followed by |
---|
822 | * a highly compressible string table.) Smaller buffer sizes give |
---|
823 | * fast adaptation but have of course the overhead of transmitting trees |
---|
824 | * more frequently. |
---|
825 | * - I can't count above 4 |
---|
826 | * The current code is general and allows DIST_BUFSIZE < LIT_BUFSIZE (to save |
---|
827 | * memory at the expense of compression). Some optimizations would be possible |
---|
828 | * if we rely on DIST_BUFSIZE == LIT_BUFSIZE. |
---|
829 | */ |
---|
830 | #define REP_3_6 16 |
---|
831 | /* repeat previous bit length 3-6 times (2 bits of repeat count) */ |
---|
832 | #define REPZ_3_10 17 |
---|
833 | /* repeat a zero length 3-10 times (3 bits of repeat count) */ |
---|
834 | #define REPZ_11_138 18 |
---|
835 | /* repeat a zero length 11-138 times (7 bits of repeat count) */ |
---|
836 | |
---|
837 | /* =========================================================================== |
---|
838 | */ |
---|
839 | /* Data structure describing a single value and its code string. */ |
---|
840 | typedef struct ct_data { |
---|
841 | union { |
---|
842 | ush freq; /* frequency count */ |
---|
843 | ush code; /* bit string */ |
---|
844 | } fc; |
---|
845 | union { |
---|
846 | ush dad; /* father node in Huffman tree */ |
---|
847 | ush len; /* length of bit string */ |
---|
848 | } dl; |
---|
849 | } ct_data; |
---|
850 | |
---|
851 | #define Freq fc.freq |
---|
852 | #define Code fc.code |
---|
853 | #define Dad dl.dad |
---|
854 | #define Len dl.len |
---|
855 | |
---|
856 | #define HEAP_SIZE (2*L_CODES + 1) |
---|
857 | /* maximum heap size */ |
---|
858 | |
---|
859 | typedef struct tree_desc { |
---|
860 | ct_data *dyn_tree; /* the dynamic tree */ |
---|
861 | ct_data *static_tree; /* corresponding static tree or NULL */ |
---|
862 | const uint8_t *extra_bits; /* extra bits for each code or NULL */ |
---|
863 | int extra_base; /* base index for extra_bits */ |
---|
864 | int elems; /* max number of elements in the tree */ |
---|
865 | int max_length; /* max bit length for the codes */ |
---|
866 | int max_code; /* largest code with non zero frequency */ |
---|
867 | } tree_desc; |
---|
868 | |
---|
869 | struct globals2 { |
---|
870 | |
---|
871 | ush heap[HEAP_SIZE]; /* heap used to build the Huffman trees */ |
---|
872 | int heap_len; /* number of elements in the heap */ |
---|
873 | int heap_max; /* element of largest frequency */ |
---|
874 | |
---|
875 | /* The sons of heap[n] are heap[2*n] and heap[2*n+1]. heap[0] is not used. |
---|
876 | * The same heap array is used to build all trees. |
---|
877 | */ |
---|
878 | |
---|
879 | ct_data dyn_ltree[HEAP_SIZE]; /* literal and length tree */ |
---|
880 | ct_data dyn_dtree[2 * D_CODES + 1]; /* distance tree */ |
---|
881 | |
---|
882 | ct_data static_ltree[L_CODES + 2]; |
---|
883 | |
---|
884 | /* The static literal tree. Since the bit lengths are imposed, there is no |
---|
885 | * need for the L_CODES extra codes used during heap construction. However |
---|
886 | * The codes 286 and 287 are needed to build a canonical tree (see ct_init |
---|
887 | * below). |
---|
888 | */ |
---|
889 | |
---|
890 | ct_data static_dtree[D_CODES]; |
---|
891 | |
---|
892 | /* The static distance tree. (Actually a trivial tree since all codes use |
---|
893 | * 5 bits.) |
---|
894 | */ |
---|
895 | |
---|
896 | ct_data bl_tree[2 * BL_CODES + 1]; |
---|
897 | |
---|
898 | /* Huffman tree for the bit lengths */ |
---|
899 | |
---|
900 | tree_desc l_desc; |
---|
901 | tree_desc d_desc; |
---|
902 | tree_desc bl_desc; |
---|
903 | |
---|
904 | ush bl_count[MAX_BITS + 1]; |
---|
905 | |
---|
906 | /* The lengths of the bit length codes are sent in order of decreasing |
---|
907 | * probability, to avoid transmitting the lengths for unused bit length codes. |
---|
908 | */ |
---|
909 | |
---|
910 | uch depth[2 * L_CODES + 1]; |
---|
911 | |
---|
912 | /* Depth of each subtree used as tie breaker for trees of equal frequency */ |
---|
913 | |
---|
914 | uch length_code[MAX_MATCH - MIN_MATCH + 1]; |
---|
915 | |
---|
916 | /* length code for each normalized match length (0 == MIN_MATCH) */ |
---|
917 | |
---|
918 | uch dist_code[512]; |
---|
919 | |
---|
920 | /* distance codes. The first 256 values correspond to the distances |
---|
921 | * 3 .. 258, the last 256 values correspond to the top 8 bits of |
---|
922 | * the 15 bit distances. |
---|
923 | */ |
---|
924 | |
---|
925 | int base_length[LENGTH_CODES]; |
---|
926 | |
---|
927 | /* First normalized length for each code (0 = MIN_MATCH) */ |
---|
928 | |
---|
929 | int base_dist[D_CODES]; |
---|
930 | |
---|
931 | /* First normalized distance for each code (0 = distance of 1) */ |
---|
932 | |
---|
933 | uch flag_buf[LIT_BUFSIZE / 8]; |
---|
934 | |
---|
935 | /* flag_buf is a bit array distinguishing literals from lengths in |
---|
936 | * l_buf, thus indicating the presence or absence of a distance. |
---|
937 | */ |
---|
938 | |
---|
939 | unsigned last_lit; /* running index in l_buf */ |
---|
940 | unsigned last_dist; /* running index in d_buf */ |
---|
941 | unsigned last_flags; /* running index in flag_buf */ |
---|
942 | uch flags; /* current flags not yet saved in flag_buf */ |
---|
943 | uch flag_bit; /* current bit used in flags */ |
---|
944 | |
---|
945 | /* bits are filled in flags starting at bit 0 (least significant). |
---|
946 | * Note: these flags are overkill in the current code since we don't |
---|
947 | * take advantage of DIST_BUFSIZE == LIT_BUFSIZE. |
---|
948 | */ |
---|
949 | |
---|
950 | ulg opt_len; /* bit length of current block with optimal trees */ |
---|
951 | ulg static_len; /* bit length of current block with static trees */ |
---|
952 | |
---|
953 | ulg compressed_len; /* total bit length of compressed file */ |
---|
954 | }; |
---|
955 | |
---|
956 | #define G2ptr ((struct globals2*)(ptr_to_globals)) |
---|
957 | #define G2 (*G2ptr) |
---|
958 | |
---|
959 | |
---|
960 | /* =========================================================================== |
---|
961 | */ |
---|
962 | static void gen_codes(ct_data * tree, int max_code); |
---|
963 | static void build_tree(tree_desc * desc); |
---|
964 | static void scan_tree(ct_data * tree, int max_code); |
---|
965 | static void send_tree(ct_data * tree, int max_code); |
---|
966 | static int build_bl_tree(void); |
---|
967 | static void send_all_trees(int lcodes, int dcodes, int blcodes); |
---|
968 | static void compress_block(ct_data * ltree, ct_data * dtree); |
---|
969 | |
---|
970 | |
---|
971 | #ifndef DEBUG |
---|
972 | /* Send a code of the given tree. c and tree must not have side effects */ |
---|
973 | # define SEND_CODE(c, tree) send_bits(tree[c].Code, tree[c].Len) |
---|
974 | #else |
---|
975 | # define SEND_CODE(c, tree) \ |
---|
976 | { \ |
---|
977 | if (verbose > 1) bb_error_msg("\ncd %3d ",(c)); \ |
---|
978 | send_bits(tree[c].Code, tree[c].Len); \ |
---|
979 | } |
---|
980 | #endif |
---|
981 | |
---|
982 | #define D_CODE(dist) \ |
---|
983 | ((dist) < 256 ? G2.dist_code[dist] : G2.dist_code[256 + ((dist)>>7)]) |
---|
984 | /* Mapping from a distance to a distance code. dist is the distance - 1 and |
---|
985 | * must not have side effects. dist_code[256] and dist_code[257] are never |
---|
986 | * used. |
---|
987 | * The arguments must not have side effects. |
---|
988 | */ |
---|
989 | |
---|
990 | |
---|
991 | /* =========================================================================== |
---|
992 | * Initialize a new block. |
---|
993 | */ |
---|
994 | static void init_block(void) |
---|
995 | { |
---|
996 | int n; /* iterates over tree elements */ |
---|
997 | |
---|
998 | /* Initialize the trees. */ |
---|
999 | for (n = 0; n < L_CODES; n++) |
---|
1000 | G2.dyn_ltree[n].Freq = 0; |
---|
1001 | for (n = 0; n < D_CODES; n++) |
---|
1002 | G2.dyn_dtree[n].Freq = 0; |
---|
1003 | for (n = 0; n < BL_CODES; n++) |
---|
1004 | G2.bl_tree[n].Freq = 0; |
---|
1005 | |
---|
1006 | G2.dyn_ltree[END_BLOCK].Freq = 1; |
---|
1007 | G2.opt_len = G2.static_len = 0; |
---|
1008 | G2.last_lit = G2.last_dist = G2.last_flags = 0; |
---|
1009 | G2.flags = 0; |
---|
1010 | G2.flag_bit = 1; |
---|
1011 | } |
---|
1012 | |
---|
1013 | |
---|
1014 | /* =========================================================================== |
---|
1015 | * Restore the heap property by moving down the tree starting at node k, |
---|
1016 | * exchanging a node with the smallest of its two sons if necessary, stopping |
---|
1017 | * when the heap property is re-established (each father smaller than its |
---|
1018 | * two sons). |
---|
1019 | */ |
---|
1020 | |
---|
1021 | /* Compares to subtrees, using the tree depth as tie breaker when |
---|
1022 | * the subtrees have equal frequency. This minimizes the worst case length. */ |
---|
1023 | #define SMALLER(tree, n, m) \ |
---|
1024 | (tree[n].Freq < tree[m].Freq \ |
---|
1025 | || (tree[n].Freq == tree[m].Freq && G2.depth[n] <= G2.depth[m])) |
---|
1026 | |
---|
1027 | static void pqdownheap(ct_data * tree, int k) |
---|
1028 | { |
---|
1029 | int v = G2.heap[k]; |
---|
1030 | int j = k << 1; /* left son of k */ |
---|
1031 | |
---|
1032 | while (j <= G2.heap_len) { |
---|
1033 | /* Set j to the smallest of the two sons: */ |
---|
1034 | if (j < G2.heap_len && SMALLER(tree, G2.heap[j + 1], G2.heap[j])) |
---|
1035 | j++; |
---|
1036 | |
---|
1037 | /* Exit if v is smaller than both sons */ |
---|
1038 | if (SMALLER(tree, v, G2.heap[j])) |
---|
1039 | break; |
---|
1040 | |
---|
1041 | /* Exchange v with the smallest son */ |
---|
1042 | G2.heap[k] = G2.heap[j]; |
---|
1043 | k = j; |
---|
1044 | |
---|
1045 | /* And continue down the tree, setting j to the left son of k */ |
---|
1046 | j <<= 1; |
---|
1047 | } |
---|
1048 | G2.heap[k] = v; |
---|
1049 | } |
---|
1050 | |
---|
1051 | |
---|
1052 | /* =========================================================================== |
---|
1053 | * Compute the optimal bit lengths for a tree and update the total bit length |
---|
1054 | * for the current block. |
---|
1055 | * IN assertion: the fields freq and dad are set, heap[heap_max] and |
---|
1056 | * above are the tree nodes sorted by increasing frequency. |
---|
1057 | * OUT assertions: the field len is set to the optimal bit length, the |
---|
1058 | * array bl_count contains the frequencies for each bit length. |
---|
1059 | * The length opt_len is updated; static_len is also updated if stree is |
---|
1060 | * not null. |
---|
1061 | */ |
---|
1062 | static void gen_bitlen(tree_desc * desc) |
---|
1063 | { |
---|
1064 | ct_data *tree = desc->dyn_tree; |
---|
1065 | const uint8_t *extra = desc->extra_bits; |
---|
1066 | int base = desc->extra_base; |
---|
1067 | int max_code = desc->max_code; |
---|
1068 | int max_length = desc->max_length; |
---|
1069 | ct_data *stree = desc->static_tree; |
---|
1070 | int h; /* heap index */ |
---|
1071 | int n, m; /* iterate over the tree elements */ |
---|
1072 | int bits; /* bit length */ |
---|
1073 | int xbits; /* extra bits */ |
---|
1074 | ush f; /* frequency */ |
---|
1075 | int overflow = 0; /* number of elements with bit length too large */ |
---|
1076 | |
---|
1077 | for (bits = 0; bits <= MAX_BITS; bits++) |
---|
1078 | G2.bl_count[bits] = 0; |
---|
1079 | |
---|
1080 | /* In a first pass, compute the optimal bit lengths (which may |
---|
1081 | * overflow in the case of the bit length tree). |
---|
1082 | */ |
---|
1083 | tree[G2.heap[G2.heap_max]].Len = 0; /* root of the heap */ |
---|
1084 | |
---|
1085 | for (h = G2.heap_max + 1; h < HEAP_SIZE; h++) { |
---|
1086 | n = G2.heap[h]; |
---|
1087 | bits = tree[tree[n].Dad].Len + 1; |
---|
1088 | if (bits > max_length) { |
---|
1089 | bits = max_length; |
---|
1090 | overflow++; |
---|
1091 | } |
---|
1092 | tree[n].Len = (ush) bits; |
---|
1093 | /* We overwrite tree[n].Dad which is no longer needed */ |
---|
1094 | |
---|
1095 | if (n > max_code) |
---|
1096 | continue; /* not a leaf node */ |
---|
1097 | |
---|
1098 | G2.bl_count[bits]++; |
---|
1099 | xbits = 0; |
---|
1100 | if (n >= base) |
---|
1101 | xbits = extra[n - base]; |
---|
1102 | f = tree[n].Freq; |
---|
1103 | G2.opt_len += (ulg) f *(bits + xbits); |
---|
1104 | |
---|
1105 | if (stree) |
---|
1106 | G2.static_len += (ulg) f * (stree[n].Len + xbits); |
---|
1107 | } |
---|
1108 | if (overflow == 0) |
---|
1109 | return; |
---|
1110 | |
---|
1111 | Trace((stderr, "\nbit length overflow\n")); |
---|
1112 | /* This happens for example on obj2 and pic of the Calgary corpus */ |
---|
1113 | |
---|
1114 | /* Find the first bit length which could increase: */ |
---|
1115 | do { |
---|
1116 | bits = max_length - 1; |
---|
1117 | while (G2.bl_count[bits] == 0) |
---|
1118 | bits--; |
---|
1119 | G2.bl_count[bits]--; /* move one leaf down the tree */ |
---|
1120 | G2.bl_count[bits + 1] += 2; /* move one overflow item as its brother */ |
---|
1121 | G2.bl_count[max_length]--; |
---|
1122 | /* The brother of the overflow item also moves one step up, |
---|
1123 | * but this does not affect bl_count[max_length] |
---|
1124 | */ |
---|
1125 | overflow -= 2; |
---|
1126 | } while (overflow > 0); |
---|
1127 | |
---|
1128 | /* Now recompute all bit lengths, scanning in increasing frequency. |
---|
1129 | * h is still equal to HEAP_SIZE. (It is simpler to reconstruct all |
---|
1130 | * lengths instead of fixing only the wrong ones. This idea is taken |
---|
1131 | * from 'ar' written by Haruhiko Okumura.) |
---|
1132 | */ |
---|
1133 | for (bits = max_length; bits != 0; bits--) { |
---|
1134 | n = G2.bl_count[bits]; |
---|
1135 | while (n != 0) { |
---|
1136 | m = G2.heap[--h]; |
---|
1137 | if (m > max_code) |
---|
1138 | continue; |
---|
1139 | if (tree[m].Len != (unsigned) bits) { |
---|
1140 | Trace((stderr, "code %d bits %d->%d\n", m, tree[m].Len, bits)); |
---|
1141 | G2.opt_len += ((int32_t) bits - tree[m].Len) * tree[m].Freq; |
---|
1142 | tree[m].Len = bits; |
---|
1143 | } |
---|
1144 | n--; |
---|
1145 | } |
---|
1146 | } |
---|
1147 | } |
---|
1148 | |
---|
1149 | |
---|
1150 | /* =========================================================================== |
---|
1151 | * Generate the codes for a given tree and bit counts (which need not be |
---|
1152 | * optimal). |
---|
1153 | * IN assertion: the array bl_count contains the bit length statistics for |
---|
1154 | * the given tree and the field len is set for all tree elements. |
---|
1155 | * OUT assertion: the field code is set for all tree elements of non |
---|
1156 | * zero code length. |
---|
1157 | */ |
---|
1158 | static void gen_codes(ct_data * tree, int max_code) |
---|
1159 | { |
---|
1160 | ush next_code[MAX_BITS + 1]; /* next code value for each bit length */ |
---|
1161 | ush code = 0; /* running code value */ |
---|
1162 | int bits; /* bit index */ |
---|
1163 | int n; /* code index */ |
---|
1164 | |
---|
1165 | /* The distribution counts are first used to generate the code values |
---|
1166 | * without bit reversal. |
---|
1167 | */ |
---|
1168 | for (bits = 1; bits <= MAX_BITS; bits++) { |
---|
1169 | next_code[bits] = code = (code + G2.bl_count[bits - 1]) << 1; |
---|
1170 | } |
---|
1171 | /* Check that the bit counts in bl_count are consistent. The last code |
---|
1172 | * must be all ones. |
---|
1173 | */ |
---|
1174 | Assert(code + G2.bl_count[MAX_BITS] - 1 == (1 << MAX_BITS) - 1, |
---|
1175 | "inconsistent bit counts"); |
---|
1176 | Tracev((stderr, "\ngen_codes: max_code %d ", max_code)); |
---|
1177 | |
---|
1178 | for (n = 0; n <= max_code; n++) { |
---|
1179 | int len = tree[n].Len; |
---|
1180 | |
---|
1181 | if (len == 0) |
---|
1182 | continue; |
---|
1183 | /* Now reverse the bits */ |
---|
1184 | tree[n].Code = bi_reverse(next_code[len]++, len); |
---|
1185 | |
---|
1186 | Tracec(tree != G2.static_ltree, |
---|
1187 | (stderr, "\nn %3d %c l %2d c %4x (%x) ", n, |
---|
1188 | (isgraph(n) ? n : ' '), len, tree[n].Code, |
---|
1189 | next_code[len] - 1)); |
---|
1190 | } |
---|
1191 | } |
---|
1192 | |
---|
1193 | |
---|
1194 | /* =========================================================================== |
---|
1195 | * Construct one Huffman tree and assigns the code bit strings and lengths. |
---|
1196 | * Update the total bit length for the current block. |
---|
1197 | * IN assertion: the field freq is set for all tree elements. |
---|
1198 | * OUT assertions: the fields len and code are set to the optimal bit length |
---|
1199 | * and corresponding code. The length opt_len is updated; static_len is |
---|
1200 | * also updated if stree is not null. The field max_code is set. |
---|
1201 | */ |
---|
1202 | |
---|
1203 | /* Remove the smallest element from the heap and recreate the heap with |
---|
1204 | * one less element. Updates heap and heap_len. */ |
---|
1205 | |
---|
1206 | #define SMALLEST 1 |
---|
1207 | /* Index within the heap array of least frequent node in the Huffman tree */ |
---|
1208 | |
---|
1209 | #define PQREMOVE(tree, top) \ |
---|
1210 | do { \ |
---|
1211 | top = G2.heap[SMALLEST]; \ |
---|
1212 | G2.heap[SMALLEST] = G2.heap[G2.heap_len--]; \ |
---|
1213 | pqdownheap(tree, SMALLEST); \ |
---|
1214 | } while (0) |
---|
1215 | |
---|
1216 | static void build_tree(tree_desc * desc) |
---|
1217 | { |
---|
1218 | ct_data *tree = desc->dyn_tree; |
---|
1219 | ct_data *stree = desc->static_tree; |
---|
1220 | int elems = desc->elems; |
---|
1221 | int n, m; /* iterate over heap elements */ |
---|
1222 | int max_code = -1; /* largest code with non zero frequency */ |
---|
1223 | int node = elems; /* next internal node of the tree */ |
---|
1224 | |
---|
1225 | /* Construct the initial heap, with least frequent element in |
---|
1226 | * heap[SMALLEST]. The sons of heap[n] are heap[2*n] and heap[2*n+1]. |
---|
1227 | * heap[0] is not used. |
---|
1228 | */ |
---|
1229 | G2.heap_len = 0; |
---|
1230 | G2.heap_max = HEAP_SIZE; |
---|
1231 | |
---|
1232 | for (n = 0; n < elems; n++) { |
---|
1233 | if (tree[n].Freq != 0) { |
---|
1234 | G2.heap[++G2.heap_len] = max_code = n; |
---|
1235 | G2.depth[n] = 0; |
---|
1236 | } else { |
---|
1237 | tree[n].Len = 0; |
---|
1238 | } |
---|
1239 | } |
---|
1240 | |
---|
1241 | /* The pkzip format requires that at least one distance code exists, |
---|
1242 | * and that at least one bit should be sent even if there is only one |
---|
1243 | * possible code. So to avoid special checks later on we force at least |
---|
1244 | * two codes of non zero frequency. |
---|
1245 | */ |
---|
1246 | while (G2.heap_len < 2) { |
---|
1247 | int new = G2.heap[++G2.heap_len] = (max_code < 2 ? ++max_code : 0); |
---|
1248 | |
---|
1249 | tree[new].Freq = 1; |
---|
1250 | G2.depth[new] = 0; |
---|
1251 | G2.opt_len--; |
---|
1252 | if (stree) |
---|
1253 | G2.static_len -= stree[new].Len; |
---|
1254 | /* new is 0 or 1 so it does not have extra bits */ |
---|
1255 | } |
---|
1256 | desc->max_code = max_code; |
---|
1257 | |
---|
1258 | /* The elements heap[heap_len/2+1 .. heap_len] are leaves of the tree, |
---|
1259 | * establish sub-heaps of increasing lengths: |
---|
1260 | */ |
---|
1261 | for (n = G2.heap_len / 2; n >= 1; n--) |
---|
1262 | pqdownheap(tree, n); |
---|
1263 | |
---|
1264 | /* Construct the Huffman tree by repeatedly combining the least two |
---|
1265 | * frequent nodes. |
---|
1266 | */ |
---|
1267 | do { |
---|
1268 | PQREMOVE(tree, n); /* n = node of least frequency */ |
---|
1269 | m = G2.heap[SMALLEST]; /* m = node of next least frequency */ |
---|
1270 | |
---|
1271 | G2.heap[--G2.heap_max] = n; /* keep the nodes sorted by frequency */ |
---|
1272 | G2.heap[--G2.heap_max] = m; |
---|
1273 | |
---|
1274 | /* Create a new node father of n and m */ |
---|
1275 | tree[node].Freq = tree[n].Freq + tree[m].Freq; |
---|
1276 | G2.depth[node] = MAX(G2.depth[n], G2.depth[m]) + 1; |
---|
1277 | tree[n].Dad = tree[m].Dad = (ush) node; |
---|
1278 | #ifdef DUMP_BL_TREE |
---|
1279 | if (tree == G2.bl_tree) { |
---|
1280 | bb_error_msg("\nnode %d(%d), sons %d(%d) %d(%d)", |
---|
1281 | node, tree[node].Freq, n, tree[n].Freq, m, tree[m].Freq); |
---|
1282 | } |
---|
1283 | #endif |
---|
1284 | /* and insert the new node in the heap */ |
---|
1285 | G2.heap[SMALLEST] = node++; |
---|
1286 | pqdownheap(tree, SMALLEST); |
---|
1287 | |
---|
1288 | } while (G2.heap_len >= 2); |
---|
1289 | |
---|
1290 | G2.heap[--G2.heap_max] = G2.heap[SMALLEST]; |
---|
1291 | |
---|
1292 | /* At this point, the fields freq and dad are set. We can now |
---|
1293 | * generate the bit lengths. |
---|
1294 | */ |
---|
1295 | gen_bitlen((tree_desc *) desc); |
---|
1296 | |
---|
1297 | /* The field len is now set, we can generate the bit codes */ |
---|
1298 | gen_codes((ct_data *) tree, max_code); |
---|
1299 | } |
---|
1300 | |
---|
1301 | |
---|
1302 | /* =========================================================================== |
---|
1303 | * Scan a literal or distance tree to determine the frequencies of the codes |
---|
1304 | * in the bit length tree. Updates opt_len to take into account the repeat |
---|
1305 | * counts. (The contribution of the bit length codes will be added later |
---|
1306 | * during the construction of bl_tree.) |
---|
1307 | */ |
---|
1308 | static void scan_tree(ct_data * tree, int max_code) |
---|
1309 | { |
---|
1310 | int n; /* iterates over all tree elements */ |
---|
1311 | int prevlen = -1; /* last emitted length */ |
---|
1312 | int curlen; /* length of current code */ |
---|
1313 | int nextlen = tree[0].Len; /* length of next code */ |
---|
1314 | int count = 0; /* repeat count of the current code */ |
---|
1315 | int max_count = 7; /* max repeat count */ |
---|
1316 | int min_count = 4; /* min repeat count */ |
---|
1317 | |
---|
1318 | if (nextlen == 0) { |
---|
1319 | max_count = 138; |
---|
1320 | min_count = 3; |
---|
1321 | } |
---|
1322 | tree[max_code + 1].Len = 0xffff; /* guard */ |
---|
1323 | |
---|
1324 | for (n = 0; n <= max_code; n++) { |
---|
1325 | curlen = nextlen; |
---|
1326 | nextlen = tree[n + 1].Len; |
---|
1327 | if (++count < max_count && curlen == nextlen) |
---|
1328 | continue; |
---|
1329 | |
---|
1330 | if (count < min_count) { |
---|
1331 | G2.bl_tree[curlen].Freq += count; |
---|
1332 | } else if (curlen != 0) { |
---|
1333 | if (curlen != prevlen) |
---|
1334 | G2.bl_tree[curlen].Freq++; |
---|
1335 | G2.bl_tree[REP_3_6].Freq++; |
---|
1336 | } else if (count <= 10) { |
---|
1337 | G2.bl_tree[REPZ_3_10].Freq++; |
---|
1338 | } else { |
---|
1339 | G2.bl_tree[REPZ_11_138].Freq++; |
---|
1340 | } |
---|
1341 | count = 0; |
---|
1342 | prevlen = curlen; |
---|
1343 | |
---|
1344 | max_count = 7; |
---|
1345 | min_count = 4; |
---|
1346 | if (nextlen == 0) { |
---|
1347 | max_count = 138; |
---|
1348 | min_count = 3; |
---|
1349 | } else if (curlen == nextlen) { |
---|
1350 | max_count = 6; |
---|
1351 | min_count = 3; |
---|
1352 | } |
---|
1353 | } |
---|
1354 | } |
---|
1355 | |
---|
1356 | |
---|
1357 | /* =========================================================================== |
---|
1358 | * Send a literal or distance tree in compressed form, using the codes in |
---|
1359 | * bl_tree. |
---|
1360 | */ |
---|
1361 | static void send_tree(ct_data * tree, int max_code) |
---|
1362 | { |
---|
1363 | int n; /* iterates over all tree elements */ |
---|
1364 | int prevlen = -1; /* last emitted length */ |
---|
1365 | int curlen; /* length of current code */ |
---|
1366 | int nextlen = tree[0].Len; /* length of next code */ |
---|
1367 | int count = 0; /* repeat count of the current code */ |
---|
1368 | int max_count = 7; /* max repeat count */ |
---|
1369 | int min_count = 4; /* min repeat count */ |
---|
1370 | |
---|
1371 | /* tree[max_code+1].Len = -1; *//* guard already set */ |
---|
1372 | if (nextlen == 0) |
---|
1373 | max_count = 138, min_count = 3; |
---|
1374 | |
---|
1375 | for (n = 0; n <= max_code; n++) { |
---|
1376 | curlen = nextlen; |
---|
1377 | nextlen = tree[n + 1].Len; |
---|
1378 | if (++count < max_count && curlen == nextlen) { |
---|
1379 | continue; |
---|
1380 | } else if (count < min_count) { |
---|
1381 | do { |
---|
1382 | SEND_CODE(curlen, G2.bl_tree); |
---|
1383 | } while (--count); |
---|
1384 | } else if (curlen != 0) { |
---|
1385 | if (curlen != prevlen) { |
---|
1386 | SEND_CODE(curlen, G2.bl_tree); |
---|
1387 | count--; |
---|
1388 | } |
---|
1389 | Assert(count >= 3 && count <= 6, " 3_6?"); |
---|
1390 | SEND_CODE(REP_3_6, G2.bl_tree); |
---|
1391 | send_bits(count - 3, 2); |
---|
1392 | } else if (count <= 10) { |
---|
1393 | SEND_CODE(REPZ_3_10, G2.bl_tree); |
---|
1394 | send_bits(count - 3, 3); |
---|
1395 | } else { |
---|
1396 | SEND_CODE(REPZ_11_138, G2.bl_tree); |
---|
1397 | send_bits(count - 11, 7); |
---|
1398 | } |
---|
1399 | count = 0; |
---|
1400 | prevlen = curlen; |
---|
1401 | if (nextlen == 0) { |
---|
1402 | max_count = 138; |
---|
1403 | min_count = 3; |
---|
1404 | } else if (curlen == nextlen) { |
---|
1405 | max_count = 6; |
---|
1406 | min_count = 3; |
---|
1407 | } else { |
---|
1408 | max_count = 7; |
---|
1409 | min_count = 4; |
---|
1410 | } |
---|
1411 | } |
---|
1412 | } |
---|
1413 | |
---|
1414 | |
---|
1415 | /* =========================================================================== |
---|
1416 | * Construct the Huffman tree for the bit lengths and return the index in |
---|
1417 | * bl_order of the last bit length code to send. |
---|
1418 | */ |
---|
1419 | static int build_bl_tree(void) |
---|
1420 | { |
---|
1421 | int max_blindex; /* index of last bit length code of non zero freq */ |
---|
1422 | |
---|
1423 | /* Determine the bit length frequencies for literal and distance trees */ |
---|
1424 | scan_tree(G2.dyn_ltree, G2.l_desc.max_code); |
---|
1425 | scan_tree(G2.dyn_dtree, G2.d_desc.max_code); |
---|
1426 | |
---|
1427 | /* Build the bit length tree: */ |
---|
1428 | build_tree(&G2.bl_desc); |
---|
1429 | /* opt_len now includes the length of the tree representations, except |
---|
1430 | * the lengths of the bit lengths codes and the 5+5+4 bits for the counts. |
---|
1431 | */ |
---|
1432 | |
---|
1433 | /* Determine the number of bit length codes to send. The pkzip format |
---|
1434 | * requires that at least 4 bit length codes be sent. (appnote.txt says |
---|
1435 | * 3 but the actual value used is 4.) |
---|
1436 | */ |
---|
1437 | for (max_blindex = BL_CODES - 1; max_blindex >= 3; max_blindex--) { |
---|
1438 | if (G2.bl_tree[bl_order[max_blindex]].Len != 0) |
---|
1439 | break; |
---|
1440 | } |
---|
1441 | /* Update opt_len to include the bit length tree and counts */ |
---|
1442 | G2.opt_len += 3 * (max_blindex + 1) + 5 + 5 + 4; |
---|
1443 | Tracev((stderr, "\ndyn trees: dyn %ld, stat %ld", G2.opt_len, G2.static_len)); |
---|
1444 | |
---|
1445 | return max_blindex; |
---|
1446 | } |
---|
1447 | |
---|
1448 | |
---|
1449 | /* =========================================================================== |
---|
1450 | * Send the header for a block using dynamic Huffman trees: the counts, the |
---|
1451 | * lengths of the bit length codes, the literal tree and the distance tree. |
---|
1452 | * IN assertion: lcodes >= 257, dcodes >= 1, blcodes >= 4. |
---|
1453 | */ |
---|
1454 | static void send_all_trees(int lcodes, int dcodes, int blcodes) |
---|
1455 | { |
---|
1456 | int rank; /* index in bl_order */ |
---|
1457 | |
---|
1458 | Assert(lcodes >= 257 && dcodes >= 1 && blcodes >= 4, "not enough codes"); |
---|
1459 | Assert(lcodes <= L_CODES && dcodes <= D_CODES |
---|
1460 | && blcodes <= BL_CODES, "too many codes"); |
---|
1461 | Tracev((stderr, "\nbl counts: ")); |
---|
1462 | send_bits(lcodes - 257, 5); /* not +255 as stated in appnote.txt */ |
---|
1463 | send_bits(dcodes - 1, 5); |
---|
1464 | send_bits(blcodes - 4, 4); /* not -3 as stated in appnote.txt */ |
---|
1465 | for (rank = 0; rank < blcodes; rank++) { |
---|
1466 | Tracev((stderr, "\nbl code %2d ", bl_order[rank])); |
---|
1467 | send_bits(G2.bl_tree[bl_order[rank]].Len, 3); |
---|
1468 | } |
---|
1469 | Tracev((stderr, "\nbl tree: sent %ld", G1.bits_sent)); |
---|
1470 | |
---|
1471 | send_tree((ct_data *) G2.dyn_ltree, lcodes - 1); /* send the literal tree */ |
---|
1472 | Tracev((stderr, "\nlit tree: sent %ld", G1.bits_sent)); |
---|
1473 | |
---|
1474 | send_tree((ct_data *) G2.dyn_dtree, dcodes - 1); /* send the distance tree */ |
---|
1475 | Tracev((stderr, "\ndist tree: sent %ld", G1.bits_sent)); |
---|
1476 | } |
---|
1477 | |
---|
1478 | |
---|
1479 | /* =========================================================================== |
---|
1480 | * Save the match info and tally the frequency counts. Return true if |
---|
1481 | * the current block must be flushed. |
---|
1482 | */ |
---|
1483 | static int ct_tally(int dist, int lc) |
---|
1484 | { |
---|
1485 | G1.l_buf[G2.last_lit++] = lc; |
---|
1486 | if (dist == 0) { |
---|
1487 | /* lc is the unmatched char */ |
---|
1488 | G2.dyn_ltree[lc].Freq++; |
---|
1489 | } else { |
---|
1490 | /* Here, lc is the match length - MIN_MATCH */ |
---|
1491 | dist--; /* dist = match distance - 1 */ |
---|
1492 | Assert((ush) dist < (ush) MAX_DIST |
---|
1493 | && (ush) lc <= (ush) (MAX_MATCH - MIN_MATCH) |
---|
1494 | && (ush) D_CODE(dist) < (ush) D_CODES, "ct_tally: bad match" |
---|
1495 | ); |
---|
1496 | |
---|
1497 | G2.dyn_ltree[G2.length_code[lc] + LITERALS + 1].Freq++; |
---|
1498 | G2.dyn_dtree[D_CODE(dist)].Freq++; |
---|
1499 | |
---|
1500 | G1.d_buf[G2.last_dist++] = dist; |
---|
1501 | G2.flags |= G2.flag_bit; |
---|
1502 | } |
---|
1503 | G2.flag_bit <<= 1; |
---|
1504 | |
---|
1505 | /* Output the flags if they fill a byte: */ |
---|
1506 | if ((G2.last_lit & 7) == 0) { |
---|
1507 | G2.flag_buf[G2.last_flags++] = G2.flags; |
---|
1508 | G2.flags = 0; |
---|
1509 | G2.flag_bit = 1; |
---|
1510 | } |
---|
1511 | /* Try to guess if it is profitable to stop the current block here */ |
---|
1512 | if ((G2.last_lit & 0xfff) == 0) { |
---|
1513 | /* Compute an upper bound for the compressed length */ |
---|
1514 | ulg out_length = G2.last_lit * 8L; |
---|
1515 | ulg in_length = (ulg) G1.strstart - G1.block_start; |
---|
1516 | int dcode; |
---|
1517 | |
---|
1518 | for (dcode = 0; dcode < D_CODES; dcode++) { |
---|
1519 | out_length += G2.dyn_dtree[dcode].Freq * (5L + extra_dbits[dcode]); |
---|
1520 | } |
---|
1521 | out_length >>= 3; |
---|
1522 | Trace((stderr, |
---|
1523 | "\nlast_lit %u, last_dist %u, in %ld, out ~%ld(%ld%%) ", |
---|
1524 | G2.last_lit, G2.last_dist, in_length, out_length, |
---|
1525 | 100L - out_length * 100L / in_length)); |
---|
1526 | if (G2.last_dist < G2.last_lit / 2 && out_length < in_length / 2) |
---|
1527 | return 1; |
---|
1528 | } |
---|
1529 | return (G2.last_lit == LIT_BUFSIZE - 1 || G2.last_dist == DIST_BUFSIZE); |
---|
1530 | /* We avoid equality with LIT_BUFSIZE because of wraparound at 64K |
---|
1531 | * on 16 bit machines and because stored blocks are restricted to |
---|
1532 | * 64K-1 bytes. |
---|
1533 | */ |
---|
1534 | } |
---|
1535 | |
---|
1536 | /* =========================================================================== |
---|
1537 | * Send the block data compressed using the given Huffman trees |
---|
1538 | */ |
---|
1539 | static void compress_block(ct_data * ltree, ct_data * dtree) |
---|
1540 | { |
---|
1541 | unsigned dist; /* distance of matched string */ |
---|
1542 | int lc; /* match length or unmatched char (if dist == 0) */ |
---|
1543 | unsigned lx = 0; /* running index in l_buf */ |
---|
1544 | unsigned dx = 0; /* running index in d_buf */ |
---|
1545 | unsigned fx = 0; /* running index in flag_buf */ |
---|
1546 | uch flag = 0; /* current flags */ |
---|
1547 | unsigned code; /* the code to send */ |
---|
1548 | int extra; /* number of extra bits to send */ |
---|
1549 | |
---|
1550 | if (G2.last_lit != 0) do { |
---|
1551 | if ((lx & 7) == 0) |
---|
1552 | flag = G2.flag_buf[fx++]; |
---|
1553 | lc = G1.l_buf[lx++]; |
---|
1554 | if ((flag & 1) == 0) { |
---|
1555 | SEND_CODE(lc, ltree); /* send a literal byte */ |
---|
1556 | Tracecv(isgraph(lc), (stderr, " '%c' ", lc)); |
---|
1557 | } else { |
---|
1558 | /* Here, lc is the match length - MIN_MATCH */ |
---|
1559 | code = G2.length_code[lc]; |
---|
1560 | SEND_CODE(code + LITERALS + 1, ltree); /* send the length code */ |
---|
1561 | extra = extra_lbits[code]; |
---|
1562 | if (extra != 0) { |
---|
1563 | lc -= G2.base_length[code]; |
---|
1564 | send_bits(lc, extra); /* send the extra length bits */ |
---|
1565 | } |
---|
1566 | dist = G1.d_buf[dx++]; |
---|
1567 | /* Here, dist is the match distance - 1 */ |
---|
1568 | code = D_CODE(dist); |
---|
1569 | Assert(code < D_CODES, "bad d_code"); |
---|
1570 | |
---|
1571 | SEND_CODE(code, dtree); /* send the distance code */ |
---|
1572 | extra = extra_dbits[code]; |
---|
1573 | if (extra != 0) { |
---|
1574 | dist -= G2.base_dist[code]; |
---|
1575 | send_bits(dist, extra); /* send the extra distance bits */ |
---|
1576 | } |
---|
1577 | } /* literal or match pair ? */ |
---|
1578 | flag >>= 1; |
---|
1579 | } while (lx < G2.last_lit); |
---|
1580 | |
---|
1581 | SEND_CODE(END_BLOCK, ltree); |
---|
1582 | } |
---|
1583 | |
---|
1584 | |
---|
1585 | /* =========================================================================== |
---|
1586 | * Determine the best encoding for the current block: dynamic trees, static |
---|
1587 | * trees or store, and output the encoded block to the zip file. This function |
---|
1588 | * returns the total compressed length for the file so far. |
---|
1589 | */ |
---|
1590 | static ulg flush_block(char *buf, ulg stored_len, int eof) |
---|
1591 | { |
---|
1592 | ulg opt_lenb, static_lenb; /* opt_len and static_len in bytes */ |
---|
1593 | int max_blindex; /* index of last bit length code of non zero freq */ |
---|
1594 | |
---|
1595 | G2.flag_buf[G2.last_flags] = G2.flags; /* Save the flags for the last 8 items */ |
---|
1596 | |
---|
1597 | /* Construct the literal and distance trees */ |
---|
1598 | build_tree(&G2.l_desc); |
---|
1599 | Tracev((stderr, "\nlit data: dyn %ld, stat %ld", G2.opt_len, G2.static_len)); |
---|
1600 | |
---|
1601 | build_tree(&G2.d_desc); |
---|
1602 | Tracev((stderr, "\ndist data: dyn %ld, stat %ld", G2.opt_len, G2.static_len)); |
---|
1603 | /* At this point, opt_len and static_len are the total bit lengths of |
---|
1604 | * the compressed block data, excluding the tree representations. |
---|
1605 | */ |
---|
1606 | |
---|
1607 | /* Build the bit length tree for the above two trees, and get the index |
---|
1608 | * in bl_order of the last bit length code to send. |
---|
1609 | */ |
---|
1610 | max_blindex = build_bl_tree(); |
---|
1611 | |
---|
1612 | /* Determine the best encoding. Compute first the block length in bytes */ |
---|
1613 | opt_lenb = (G2.opt_len + 3 + 7) >> 3; |
---|
1614 | static_lenb = (G2.static_len + 3 + 7) >> 3; |
---|
1615 | |
---|
1616 | Trace((stderr, |
---|
1617 | "\nopt %lu(%lu) stat %lu(%lu) stored %lu lit %u dist %u ", |
---|
1618 | opt_lenb, G2.opt_len, static_lenb, G2.static_len, stored_len, |
---|
1619 | G2.last_lit, G2.last_dist)); |
---|
1620 | |
---|
1621 | if (static_lenb <= opt_lenb) |
---|
1622 | opt_lenb = static_lenb; |
---|
1623 | |
---|
1624 | /* If compression failed and this is the first and last block, |
---|
1625 | * and if the zip file can be seeked (to rewrite the local header), |
---|
1626 | * the whole file is transformed into a stored file: |
---|
1627 | */ |
---|
1628 | if (stored_len <= opt_lenb && eof && G2.compressed_len == 0L && seekable()) { |
---|
1629 | /* Since LIT_BUFSIZE <= 2*WSIZE, the input data must be there: */ |
---|
1630 | if (buf == NULL) |
---|
1631 | bb_error_msg("block vanished"); |
---|
1632 | |
---|
1633 | copy_block(buf, (unsigned) stored_len, 0); /* without header */ |
---|
1634 | G2.compressed_len = stored_len << 3; |
---|
1635 | |
---|
1636 | } else if (stored_len + 4 <= opt_lenb && buf != NULL) { |
---|
1637 | /* 4: two words for the lengths */ |
---|
1638 | /* The test buf != NULL is only necessary if LIT_BUFSIZE > WSIZE. |
---|
1639 | * Otherwise we can't have processed more than WSIZE input bytes since |
---|
1640 | * the last block flush, because compression would have been |
---|
1641 | * successful. If LIT_BUFSIZE <= WSIZE, it is never too late to |
---|
1642 | * transform a block into a stored block. |
---|
1643 | */ |
---|
1644 | send_bits((STORED_BLOCK << 1) + eof, 3); /* send block type */ |
---|
1645 | G2.compressed_len = (G2.compressed_len + 3 + 7) & ~7L; |
---|
1646 | G2.compressed_len += (stored_len + 4) << 3; |
---|
1647 | |
---|
1648 | copy_block(buf, (unsigned) stored_len, 1); /* with header */ |
---|
1649 | |
---|
1650 | } else if (static_lenb == opt_lenb) { |
---|
1651 | send_bits((STATIC_TREES << 1) + eof, 3); |
---|
1652 | compress_block((ct_data *) G2.static_ltree, (ct_data *) G2.static_dtree); |
---|
1653 | G2.compressed_len += 3 + G2.static_len; |
---|
1654 | } else { |
---|
1655 | send_bits((DYN_TREES << 1) + eof, 3); |
---|
1656 | send_all_trees(G2.l_desc.max_code + 1, G2.d_desc.max_code + 1, |
---|
1657 | max_blindex + 1); |
---|
1658 | compress_block((ct_data *) G2.dyn_ltree, (ct_data *) G2.dyn_dtree); |
---|
1659 | G2.compressed_len += 3 + G2.opt_len; |
---|
1660 | } |
---|
1661 | Assert(G2.compressed_len == G1.bits_sent, "bad compressed size"); |
---|
1662 | init_block(); |
---|
1663 | |
---|
1664 | if (eof) { |
---|
1665 | bi_windup(); |
---|
1666 | G2.compressed_len += 7; /* align on byte boundary */ |
---|
1667 | } |
---|
1668 | Tracev((stderr, "\ncomprlen %lu(%lu) ", G2.compressed_len >> 3, |
---|
1669 | G2.compressed_len - 7 * eof)); |
---|
1670 | |
---|
1671 | return G2.compressed_len >> 3; |
---|
1672 | } |
---|
1673 | |
---|
1674 | |
---|
1675 | /* =========================================================================== |
---|
1676 | * Update a hash value with the given input byte |
---|
1677 | * IN assertion: all calls to to UPDATE_HASH are made with consecutive |
---|
1678 | * input characters, so that a running hash key can be computed from the |
---|
1679 | * previous key instead of complete recalculation each time. |
---|
1680 | */ |
---|
1681 | #define UPDATE_HASH(h, c) (h = (((h)<<H_SHIFT) ^ (c)) & HASH_MASK) |
---|
1682 | |
---|
1683 | |
---|
1684 | /* =========================================================================== |
---|
1685 | * Same as above, but achieves better compression. We use a lazy |
---|
1686 | * evaluation for matches: a match is finally adopted only if there is |
---|
1687 | * no better match at the next window position. |
---|
1688 | * |
---|
1689 | * Processes a new input file and return its compressed length. Sets |
---|
1690 | * the compressed length, crc, deflate flags and internal file |
---|
1691 | * attributes. |
---|
1692 | */ |
---|
1693 | |
---|
1694 | /* Flush the current block, with given end-of-file flag. |
---|
1695 | * IN assertion: strstart is set to the end of the current match. */ |
---|
1696 | #define FLUSH_BLOCK(eof) \ |
---|
1697 | flush_block( \ |
---|
1698 | G1.block_start >= 0L \ |
---|
1699 | ? (char*)&G1.window[(unsigned)G1.block_start] \ |
---|
1700 | : (char*)NULL, \ |
---|
1701 | (ulg)G1.strstart - G1.block_start, \ |
---|
1702 | (eof) \ |
---|
1703 | ) |
---|
1704 | |
---|
1705 | /* Insert string s in the dictionary and set match_head to the previous head |
---|
1706 | * of the hash chain (the most recent string with same hash key). Return |
---|
1707 | * the previous length of the hash chain. |
---|
1708 | * IN assertion: all calls to to INSERT_STRING are made with consecutive |
---|
1709 | * input characters and the first MIN_MATCH bytes of s are valid |
---|
1710 | * (except for the last MIN_MATCH-1 bytes of the input file). */ |
---|
1711 | #define INSERT_STRING(s, match_head) \ |
---|
1712 | do { \ |
---|
1713 | UPDATE_HASH(G1.ins_h, G1.window[(s) + MIN_MATCH-1]); \ |
---|
1714 | G1.prev[(s) & WMASK] = match_head = head[G1.ins_h]; \ |
---|
1715 | head[G1.ins_h] = (s); \ |
---|
1716 | } while (0) |
---|
1717 | |
---|
1718 | static ulg deflate(void) |
---|
1719 | { |
---|
1720 | IPos hash_head; /* head of hash chain */ |
---|
1721 | IPos prev_match; /* previous match */ |
---|
1722 | int flush; /* set if current block must be flushed */ |
---|
1723 | int match_available = 0; /* set if previous match exists */ |
---|
1724 | unsigned match_length = MIN_MATCH - 1; /* length of best match */ |
---|
1725 | |
---|
1726 | /* Process the input block. */ |
---|
1727 | while (G1.lookahead != 0) { |
---|
1728 | /* Insert the string window[strstart .. strstart+2] in the |
---|
1729 | * dictionary, and set hash_head to the head of the hash chain: |
---|
1730 | */ |
---|
1731 | INSERT_STRING(G1.strstart, hash_head); |
---|
1732 | |
---|
1733 | /* Find the longest match, discarding those <= prev_length. |
---|
1734 | */ |
---|
1735 | G1.prev_length = match_length; |
---|
1736 | prev_match = G1.match_start; |
---|
1737 | match_length = MIN_MATCH - 1; |
---|
1738 | |
---|
1739 | if (hash_head != 0 && G1.prev_length < max_lazy_match |
---|
1740 | && G1.strstart - hash_head <= MAX_DIST |
---|
1741 | ) { |
---|
1742 | /* To simplify the code, we prevent matches with the string |
---|
1743 | * of window index 0 (in particular we have to avoid a match |
---|
1744 | * of the string with itself at the start of the input file). |
---|
1745 | */ |
---|
1746 | match_length = longest_match(hash_head); |
---|
1747 | /* longest_match() sets match_start */ |
---|
1748 | if (match_length > G1.lookahead) |
---|
1749 | match_length = G1.lookahead; |
---|
1750 | |
---|
1751 | /* Ignore a length 3 match if it is too distant: */ |
---|
1752 | if (match_length == MIN_MATCH && G1.strstart - G1.match_start > TOO_FAR) { |
---|
1753 | /* If prev_match is also MIN_MATCH, G1.match_start is garbage |
---|
1754 | * but we will ignore the current match anyway. |
---|
1755 | */ |
---|
1756 | match_length--; |
---|
1757 | } |
---|
1758 | } |
---|
1759 | /* If there was a match at the previous step and the current |
---|
1760 | * match is not better, output the previous match: |
---|
1761 | */ |
---|
1762 | if (G1.prev_length >= MIN_MATCH && match_length <= G1.prev_length) { |
---|
1763 | check_match(G1.strstart - 1, prev_match, G1.prev_length); |
---|
1764 | flush = ct_tally(G1.strstart - 1 - prev_match, G1.prev_length - MIN_MATCH); |
---|
1765 | |
---|
1766 | /* Insert in hash table all strings up to the end of the match. |
---|
1767 | * strstart-1 and strstart are already inserted. |
---|
1768 | */ |
---|
1769 | G1.lookahead -= G1.prev_length - 1; |
---|
1770 | G1.prev_length -= 2; |
---|
1771 | do { |
---|
1772 | G1.strstart++; |
---|
1773 | INSERT_STRING(G1.strstart, hash_head); |
---|
1774 | /* strstart never exceeds WSIZE-MAX_MATCH, so there are |
---|
1775 | * always MIN_MATCH bytes ahead. If lookahead < MIN_MATCH |
---|
1776 | * these bytes are garbage, but it does not matter since the |
---|
1777 | * next lookahead bytes will always be emitted as literals. |
---|
1778 | */ |
---|
1779 | } while (--G1.prev_length != 0); |
---|
1780 | match_available = 0; |
---|
1781 | match_length = MIN_MATCH - 1; |
---|
1782 | G1.strstart++; |
---|
1783 | if (flush) { |
---|
1784 | FLUSH_BLOCK(0); |
---|
1785 | G1.block_start = G1.strstart; |
---|
1786 | } |
---|
1787 | } else if (match_available) { |
---|
1788 | /* If there was no match at the previous position, output a |
---|
1789 | * single literal. If there was a match but the current match |
---|
1790 | * is longer, truncate the previous match to a single literal. |
---|
1791 | */ |
---|
1792 | Tracevv((stderr, "%c", G1.window[G1.strstart - 1])); |
---|
1793 | if (ct_tally(0, G1.window[G1.strstart - 1])) { |
---|
1794 | FLUSH_BLOCK(0); |
---|
1795 | G1.block_start = G1.strstart; |
---|
1796 | } |
---|
1797 | G1.strstart++; |
---|
1798 | G1.lookahead--; |
---|
1799 | } else { |
---|
1800 | /* There is no previous match to compare with, wait for |
---|
1801 | * the next step to decide. |
---|
1802 | */ |
---|
1803 | match_available = 1; |
---|
1804 | G1.strstart++; |
---|
1805 | G1.lookahead--; |
---|
1806 | } |
---|
1807 | Assert(G1.strstart <= G1.isize && lookahead <= G1.isize, "a bit too far"); |
---|
1808 | |
---|
1809 | /* Make sure that we always have enough lookahead, except |
---|
1810 | * at the end of the input file. We need MAX_MATCH bytes |
---|
1811 | * for the next match, plus MIN_MATCH bytes to insert the |
---|
1812 | * string following the next match. |
---|
1813 | */ |
---|
1814 | while (G1.lookahead < MIN_LOOKAHEAD && !G1.eofile) |
---|
1815 | fill_window(); |
---|
1816 | } |
---|
1817 | if (match_available) |
---|
1818 | ct_tally(0, G1.window[G1.strstart - 1]); |
---|
1819 | |
---|
1820 | return FLUSH_BLOCK(1); /* eof */ |
---|
1821 | } |
---|
1822 | |
---|
1823 | |
---|
1824 | /* =========================================================================== |
---|
1825 | * Initialize the bit string routines. |
---|
1826 | */ |
---|
1827 | static void bi_init(void) |
---|
1828 | { |
---|
1829 | G1.bi_buf = 0; |
---|
1830 | G1.bi_valid = 0; |
---|
1831 | #ifdef DEBUG |
---|
1832 | G1.bits_sent = 0L; |
---|
1833 | #endif |
---|
1834 | } |
---|
1835 | |
---|
1836 | |
---|
1837 | /* =========================================================================== |
---|
1838 | * Initialize the "longest match" routines for a new file |
---|
1839 | */ |
---|
1840 | static void lm_init(ush * flagsp) |
---|
1841 | { |
---|
1842 | unsigned j; |
---|
1843 | |
---|
1844 | /* Initialize the hash table. */ |
---|
1845 | memset(head, 0, HASH_SIZE * sizeof(*head)); |
---|
1846 | /* prev will be initialized on the fly */ |
---|
1847 | |
---|
1848 | /* speed options for the general purpose bit flag */ |
---|
1849 | *flagsp |= 2; /* FAST 4, SLOW 2 */ |
---|
1850 | /* ??? reduce max_chain_length for binary files */ |
---|
1851 | |
---|
1852 | G1.strstart = 0; |
---|
1853 | G1.block_start = 0L; |
---|
1854 | |
---|
1855 | G1.lookahead = file_read(G1.window, |
---|
1856 | sizeof(int) <= 2 ? (unsigned) WSIZE : 2 * WSIZE); |
---|
1857 | |
---|
1858 | if (G1.lookahead == 0 || G1.lookahead == (unsigned) -1) { |
---|
1859 | G1.eofile = 1; |
---|
1860 | G1.lookahead = 0; |
---|
1861 | return; |
---|
1862 | } |
---|
1863 | G1.eofile = 0; |
---|
1864 | /* Make sure that we always have enough lookahead. This is important |
---|
1865 | * if input comes from a device such as a tty. |
---|
1866 | */ |
---|
1867 | while (G1.lookahead < MIN_LOOKAHEAD && !G1.eofile) |
---|
1868 | fill_window(); |
---|
1869 | |
---|
1870 | G1.ins_h = 0; |
---|
1871 | for (j = 0; j < MIN_MATCH - 1; j++) |
---|
1872 | UPDATE_HASH(G1.ins_h, G1.window[j]); |
---|
1873 | /* If lookahead < MIN_MATCH, ins_h is garbage, but this is |
---|
1874 | * not important since only literal bytes will be emitted. |
---|
1875 | */ |
---|
1876 | } |
---|
1877 | |
---|
1878 | |
---|
1879 | /* =========================================================================== |
---|
1880 | * Allocate the match buffer, initialize the various tables and save the |
---|
1881 | * location of the internal file attribute (ascii/binary) and method |
---|
1882 | * (DEFLATE/STORE). |
---|
1883 | * One callsite in zip() |
---|
1884 | */ |
---|
1885 | static void ct_init(void) |
---|
1886 | { |
---|
1887 | int n; /* iterates over tree elements */ |
---|
1888 | int length; /* length value */ |
---|
1889 | int code; /* code value */ |
---|
1890 | int dist; /* distance index */ |
---|
1891 | |
---|
1892 | G2.compressed_len = 0L; |
---|
1893 | |
---|
1894 | #ifdef NOT_NEEDED |
---|
1895 | if (G2.static_dtree[0].Len != 0) |
---|
1896 | return; /* ct_init already called */ |
---|
1897 | #endif |
---|
1898 | |
---|
1899 | /* Initialize the mapping length (0..255) -> length code (0..28) */ |
---|
1900 | length = 0; |
---|
1901 | for (code = 0; code < LENGTH_CODES - 1; code++) { |
---|
1902 | G2.base_length[code] = length; |
---|
1903 | for (n = 0; n < (1 << extra_lbits[code]); n++) { |
---|
1904 | G2.length_code[length++] = code; |
---|
1905 | } |
---|
1906 | } |
---|
1907 | Assert(length == 256, "ct_init: length != 256"); |
---|
1908 | /* Note that the length 255 (match length 258) can be represented |
---|
1909 | * in two different ways: code 284 + 5 bits or code 285, so we |
---|
1910 | * overwrite length_code[255] to use the best encoding: |
---|
1911 | */ |
---|
1912 | G2.length_code[length - 1] = code; |
---|
1913 | |
---|
1914 | /* Initialize the mapping dist (0..32K) -> dist code (0..29) */ |
---|
1915 | dist = 0; |
---|
1916 | for (code = 0; code < 16; code++) { |
---|
1917 | G2.base_dist[code] = dist; |
---|
1918 | for (n = 0; n < (1 << extra_dbits[code]); n++) { |
---|
1919 | G2.dist_code[dist++] = code; |
---|
1920 | } |
---|
1921 | } |
---|
1922 | Assert(dist == 256, "ct_init: dist != 256"); |
---|
1923 | dist >>= 7; /* from now on, all distances are divided by 128 */ |
---|
1924 | for (; code < D_CODES; code++) { |
---|
1925 | G2.base_dist[code] = dist << 7; |
---|
1926 | for (n = 0; n < (1 << (extra_dbits[code] - 7)); n++) { |
---|
1927 | G2.dist_code[256 + dist++] = code; |
---|
1928 | } |
---|
1929 | } |
---|
1930 | Assert(dist == 256, "ct_init: 256+dist != 512"); |
---|
1931 | |
---|
1932 | /* Construct the codes of the static literal tree */ |
---|
1933 | /* already zeroed - it's in bss |
---|
1934 | for (n = 0; n <= MAX_BITS; n++) |
---|
1935 | G2.bl_count[n] = 0; */ |
---|
1936 | |
---|
1937 | n = 0; |
---|
1938 | while (n <= 143) { |
---|
1939 | G2.static_ltree[n++].Len = 8; |
---|
1940 | G2.bl_count[8]++; |
---|
1941 | } |
---|
1942 | while (n <= 255) { |
---|
1943 | G2.static_ltree[n++].Len = 9; |
---|
1944 | G2.bl_count[9]++; |
---|
1945 | } |
---|
1946 | while (n <= 279) { |
---|
1947 | G2.static_ltree[n++].Len = 7; |
---|
1948 | G2.bl_count[7]++; |
---|
1949 | } |
---|
1950 | while (n <= 287) { |
---|
1951 | G2.static_ltree[n++].Len = 8; |
---|
1952 | G2.bl_count[8]++; |
---|
1953 | } |
---|
1954 | /* Codes 286 and 287 do not exist, but we must include them in the |
---|
1955 | * tree construction to get a canonical Huffman tree (longest code |
---|
1956 | * all ones) |
---|
1957 | */ |
---|
1958 | gen_codes((ct_data *) G2.static_ltree, L_CODES + 1); |
---|
1959 | |
---|
1960 | /* The static distance tree is trivial: */ |
---|
1961 | for (n = 0; n < D_CODES; n++) { |
---|
1962 | G2.static_dtree[n].Len = 5; |
---|
1963 | G2.static_dtree[n].Code = bi_reverse(n, 5); |
---|
1964 | } |
---|
1965 | |
---|
1966 | /* Initialize the first block of the first file: */ |
---|
1967 | init_block(); |
---|
1968 | } |
---|
1969 | |
---|
1970 | |
---|
1971 | /* =========================================================================== |
---|
1972 | * Deflate in to out. |
---|
1973 | * IN assertions: the input and output buffers are cleared. |
---|
1974 | */ |
---|
1975 | |
---|
1976 | static void zip(ulg time_stamp) |
---|
1977 | { |
---|
1978 | ush deflate_flags = 0; /* pkzip -es, -en or -ex equivalent */ |
---|
1979 | |
---|
1980 | G1.outcnt = 0; |
---|
1981 | |
---|
1982 | /* Write the header to the gzip file. See algorithm.doc for the format */ |
---|
1983 | /* magic header for gzip files: 1F 8B */ |
---|
1984 | /* compression method: 8 (DEFLATED) */ |
---|
1985 | /* general flags: 0 */ |
---|
1986 | put_32bit(0x00088b1f); |
---|
1987 | put_32bit(time_stamp); |
---|
1988 | |
---|
1989 | /* Write deflated file to zip file */ |
---|
1990 | G1.crc = ~0; |
---|
1991 | |
---|
1992 | bi_init(); |
---|
1993 | ct_init(); |
---|
1994 | lm_init(&deflate_flags); |
---|
1995 | |
---|
1996 | put_8bit(deflate_flags); /* extra flags */ |
---|
1997 | put_8bit(3); /* OS identifier = 3 (Unix) */ |
---|
1998 | |
---|
1999 | deflate(); |
---|
2000 | |
---|
2001 | /* Write the crc and uncompressed size */ |
---|
2002 | put_32bit(~G1.crc); |
---|
2003 | put_32bit(G1.isize); |
---|
2004 | |
---|
2005 | flush_outbuf(); |
---|
2006 | } |
---|
2007 | |
---|
2008 | |
---|
2009 | /* ======================================================================== */ |
---|
2010 | static |
---|
2011 | char* make_new_name_gzip(char *filename) |
---|
2012 | { |
---|
2013 | return xasprintf("%s.gz", filename); |
---|
2014 | } |
---|
2015 | |
---|
2016 | static |
---|
2017 | USE_DESKTOP(long long) int pack_gzip(void) |
---|
2018 | { |
---|
2019 | struct stat s; |
---|
2020 | |
---|
2021 | clear_bufs(); |
---|
2022 | s.st_ctime = 0; |
---|
2023 | fstat(STDIN_FILENO, &s); |
---|
2024 | zip(s.st_ctime); |
---|
2025 | return 0; |
---|
2026 | } |
---|
2027 | |
---|
2028 | int gzip_main(int argc, char **argv); |
---|
2029 | int gzip_main(int argc, char **argv) |
---|
2030 | { |
---|
2031 | unsigned opt; |
---|
2032 | |
---|
2033 | /* Must match bbunzip's constants OPT_STDOUT, OPT_FORCE! */ |
---|
2034 | opt = getopt32(argv, "cfv" USE_GUNZIP("d") "q123456789" ); |
---|
2035 | option_mask32 &= 0x7; /* Clear -d, ignore -q, -0..9 */ |
---|
2036 | //if (opt & 0x1) // -c |
---|
2037 | //if (opt & 0x2) // -f |
---|
2038 | //if (opt & 0x4) // -v |
---|
2039 | #if ENABLE_GUNZIP /* gunzip_main may not be visible... */ |
---|
2040 | if (opt & 0x8) { // -d |
---|
2041 | return gunzip_main(argc, argv); |
---|
2042 | } |
---|
2043 | #endif |
---|
2044 | argv += optind; |
---|
2045 | |
---|
2046 | PTR_TO_GLOBALS = xzalloc(sizeof(struct globals) + sizeof(struct globals2)) |
---|
2047 | + sizeof(struct globals); |
---|
2048 | G2.l_desc.dyn_tree = G2.dyn_ltree; |
---|
2049 | G2.l_desc.static_tree = G2.static_ltree; |
---|
2050 | G2.l_desc.extra_bits = extra_lbits; |
---|
2051 | G2.l_desc.extra_base = LITERALS + 1; |
---|
2052 | G2.l_desc.elems = L_CODES; |
---|
2053 | G2.l_desc.max_length = MAX_BITS; |
---|
2054 | //G2.l_desc.max_code = 0; |
---|
2055 | |
---|
2056 | G2.d_desc.dyn_tree = G2.dyn_dtree; |
---|
2057 | G2.d_desc.static_tree = G2.static_dtree; |
---|
2058 | G2.d_desc.extra_bits = extra_dbits; |
---|
2059 | //G2.d_desc.extra_base = 0; |
---|
2060 | G2.d_desc.elems = D_CODES; |
---|
2061 | G2.d_desc.max_length = MAX_BITS; |
---|
2062 | //G2.d_desc.max_code = 0; |
---|
2063 | |
---|
2064 | G2.bl_desc.dyn_tree = G2.bl_tree; |
---|
2065 | //G2.bl_desc.static_tree = NULL; |
---|
2066 | G2.bl_desc.extra_bits = extra_blbits, |
---|
2067 | //G2.bl_desc.extra_base = 0; |
---|
2068 | G2.bl_desc.elems = BL_CODES; |
---|
2069 | G2.bl_desc.max_length = MAX_BL_BITS; |
---|
2070 | //G2.bl_desc.max_code = 0; |
---|
2071 | |
---|
2072 | /* Allocate all global buffers (for DYN_ALLOC option) */ |
---|
2073 | ALLOC(uch, G1.l_buf, INBUFSIZ); |
---|
2074 | ALLOC(uch, G1.outbuf, OUTBUFSIZ); |
---|
2075 | ALLOC(ush, G1.d_buf, DIST_BUFSIZE); |
---|
2076 | ALLOC(uch, G1.window, 2L * WSIZE); |
---|
2077 | ALLOC(ush, G1.prev, 1L << BITS); |
---|
2078 | |
---|
2079 | /* Initialise the CRC32 table */ |
---|
2080 | G1.crc_32_tab = crc32_filltable(NULL, 0); |
---|
2081 | |
---|
2082 | return bbunpack(argv, make_new_name_gzip, pack_gzip); |
---|
2083 | } |
---|