blob: 44d894244d75743788421286da351b278a2db0c6 [file] [log] [blame]
"Robert P. J. Day"63fc1a92006-07-02 19:47:05 +00001/* vi: set sw=4 ts=4: */
Robert Griebl94a6a952002-05-19 19:00:14 +00002/* uncompress for busybox -- (c) 2002 Robert Griebl
3 *
Eric Andersenc7bda1c2004-03-15 08:29:22 +00004 * based on the original compress42.c source
Robert Griebl94a6a952002-05-19 19:00:14 +00005 * (see disclaimer below)
6 */
7
Robert Griebl94a6a952002-05-19 19:00:14 +00008/* (N)compress42.c - File compression ala IEEE Computer, Mar 1992.
9 *
10 * Authors:
11 * Spencer W. Thomas (decvax!harpo!utah-cs!utah-gr!thomas)
12 * Jim McKie (decvax!mcvax!jim)
13 * Steve Davies (decvax!vax135!petsd!peora!srd)
14 * Ken Turkowski (decvax!decwrl!turtlevax!ken)
15 * James A. Woods (decvax!ihnp4!ames!jaw)
16 * Joe Orost (decvax!vax135!petsd!joe)
17 * Dave Mack (csu@alembic.acs.com)
18 * Peter Jannesen, Network Communication Systems
19 * (peter@ncs.nl)
20 *
21 * marc@suse.de : a small security fix for a buffer overflow
22 *
23 * [... History snipped ...]
24 *
25 */
Robert Griebl94a6a952002-05-19 19:00:14 +000026
Denis Vlasenkoe9ad84d2008-08-05 13:10:34 +000027#include "libbb.h"
Denys Vlasenko833d4e72010-11-03 02:38:31 +010028#include "archive.h"
Denis Vlasenkoe9ad84d2008-08-05 13:10:34 +000029
30
Eric Andersenaff114c2004-04-14 17:51:38 +000031/* Default input buffer size */
Denys Vlasenkoe4dcba12010-10-28 18:57:19 +020032#define IBUFSIZ 2048
Robert Griebl94a6a952002-05-19 19:00:14 +000033
Glenn L McGrathfedbfe42002-11-28 09:09:47 +000034/* Default output buffer size */
Denys Vlasenkoe4dcba12010-10-28 18:57:19 +020035#define OBUFSIZ 2048
Robert Griebl94a6a952002-05-19 19:00:14 +000036
Glenn L McGrathfedbfe42002-11-28 09:09:47 +000037/* Defines for third byte of header */
Denis Vlasenkoe8a07882007-06-10 15:08:44 +000038#define BIT_MASK 0x1f /* Mask for 'number of compresssion bits' */
39 /* Masks 0x20 and 0x40 are free. */
40 /* I think 0x20 should mean that there is */
41 /* a fourth header byte (for expansion). */
42#define BLOCK_MODE 0x80 /* Block compression if table is full and */
43 /* compression rate is dropping flush tables */
44 /* the next two codes should not be changed lightly, as they must not */
45 /* lie within the contiguous general code space. */
46#define FIRST 257 /* first free entry */
47#define CLEAR 256 /* table clear output code */
Robert Griebl94a6a952002-05-19 19:00:14 +000048
Denis Vlasenkoe8a07882007-06-10 15:08:44 +000049#define INIT_BITS 9 /* initial number of bits/code */
Robert Griebl94a6a952002-05-19 19:00:14 +000050
51
Glenn L McGrathfedbfe42002-11-28 09:09:47 +000052/* machine variants which require cc -Dmachine: pdp11, z8000, DOS */
Denis Vlasenkoe8a07882007-06-10 15:08:44 +000053#define HBITS 17 /* 50% occupancy */
54#define HSIZE (1<<HBITS)
55#define HMASK (HSIZE-1) /* unused */
56#define HPRIME 9941 /* unused */
57#define BITS 16
58#define BITS_STR "16"
59#undef MAXSEG_64K /* unused */
60#define MAXCODE(n) (1L << (n))
Robert Griebl94a6a952002-05-19 19:00:14 +000061
Denis Vlasenkoe8a07882007-06-10 15:08:44 +000062#define htabof(i) htab[i]
63#define codetabof(i) codetab[i]
64#define tab_prefixof(i) codetabof(i)
65#define tab_suffixof(i) ((unsigned char *)(htab))[i]
66#define de_stack ((unsigned char *)&(htab[HSIZE-1]))
67#define clear_tab_prefixof() memset(codetab, 0, 256)
Robert Griebl94a6a952002-05-19 19:00:14 +000068
Robert Griebl94a6a952002-05-19 19:00:14 +000069/*
70 * Decompress stdin to stdout. This routine adapts to the codes in the
71 * file building the "string" table on-the-fly; requiring no table to
Denis Vlasenkoe8a07882007-06-10 15:08:44 +000072 * be stored in the compressed file.
Robert Griebl94a6a952002-05-19 19:00:14 +000073 */
74
Denis Vlasenko5e34ff22009-04-21 11:09:40 +000075IF_DESKTOP(long long) int FAST_FUNC
Denis Vlasenkoe9ad84d2008-08-05 13:10:34 +000076unpack_Z_stream(int fd_in, int fd_out)
Robert Griebl94a6a952002-05-19 19:00:14 +000077{
Denis Vlasenko5e34ff22009-04-21 11:09:40 +000078 IF_DESKTOP(long long total_written = 0;)
79 IF_DESKTOP(long long) int retval = -1;
Glenn L McGrathfedbfe42002-11-28 09:09:47 +000080 unsigned char *stackp;
Denis Vlasenko87468852007-04-13 23:22:00 +000081 long code;
Glenn L McGrathfedbfe42002-11-28 09:09:47 +000082 int finchar;
Denis Vlasenko87468852007-04-13 23:22:00 +000083 long oldcode;
84 long incode;
Glenn L McGrathfedbfe42002-11-28 09:09:47 +000085 int inbits;
86 int posbits;
87 int outpos;
88 int insize;
89 int bitmask;
Denis Vlasenko87468852007-04-13 23:22:00 +000090 long free_ent;
91 long maxcode;
92 long maxmaxcode;
Glenn L McGrathfedbfe42002-11-28 09:09:47 +000093 int n_bits;
94 int rsize = 0;
Denis Vlasenkoe8a07882007-06-10 15:08:44 +000095 unsigned char *inbuf; /* were eating insane amounts of stack - */
96 unsigned char *outbuf; /* bad for some embedded targets */
97 unsigned char *htab;
98 unsigned short *codetab;
Denis Vlasenko447b5432007-01-05 19:49:02 +000099
100 /* Hmm, these were statics - why?! */
101 /* user settable max # bits/code */
102 int maxbits; /* = BITS; */
103 /* block compress mode -C compatible with 2.0 */
104 int block_mode; /* = BLOCK_MODE; */
105
Denis Vlasenkoe8a07882007-06-10 15:08:44 +0000106 inbuf = xzalloc(IBUFSIZ + 64);
107 outbuf = xzalloc(OBUFSIZ + 2048);
108 htab = xzalloc(HSIZE); /* wsn't zeroed out before, maybe can xmalloc? */
109 codetab = xzalloc(HSIZE * sizeof(codetab[0]));
Robert Griebl94a6a952002-05-19 19:00:14 +0000110
111 insize = 0;
112
Denis Vlasenko447b5432007-01-05 19:49:02 +0000113 /* xread isn't good here, we have to return - caller may want
114 * to do some cleanup (e.g. delete incomplete unpacked file etc) */
115 if (full_read(fd_in, inbuf, 1) != 1) {
116 bb_error_msg("short read");
Denis Vlasenkoe8a07882007-06-10 15:08:44 +0000117 goto err;
Denis Vlasenko447b5432007-01-05 19:49:02 +0000118 }
Robert Griebl94a6a952002-05-19 19:00:14 +0000119
120 maxbits = inbuf[0] & BIT_MASK;
121 block_mode = inbuf[0] & BLOCK_MODE;
122 maxmaxcode = MAXCODE(maxbits);
123
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000124 if (maxbits > BITS) {
Denis Vlasenko447b5432007-01-05 19:49:02 +0000125 bb_error_msg("compressed with %d bits, can only handle "
Denis Vlasenkoe8a07882007-06-10 15:08:44 +0000126 BITS_STR" bits", maxbits);
127 goto err;
Robert Griebl94a6a952002-05-19 19:00:14 +0000128 }
129
Denis Vlasenko447b5432007-01-05 19:49:02 +0000130 n_bits = INIT_BITS;
131 maxcode = MAXCODE(INIT_BITS) - 1;
132 bitmask = (1 << INIT_BITS) - 1;
Robert Griebl94a6a952002-05-19 19:00:14 +0000133 oldcode = -1;
134 finchar = 0;
135 outpos = 0;
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000136 posbits = 0 << 3;
Robert Griebl94a6a952002-05-19 19:00:14 +0000137
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000138 free_ent = ((block_mode) ? FIRST : 256);
Robert Griebl94a6a952002-05-19 19:00:14 +0000139
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000140 /* As above, initialize the first 256 entries in the table. */
Denis Vlasenkoe8a07882007-06-10 15:08:44 +0000141 /*clear_tab_prefixof(); - done by xzalloc */
Robert Griebl94a6a952002-05-19 19:00:14 +0000142
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000143 for (code = 255; code >= 0; --code) {
144 tab_suffixof(code) = (unsigned char) code;
145 }
Robert Griebl94a6a952002-05-19 19:00:14 +0000146
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000147 do {
Denis Vlasenko447b5432007-01-05 19:49:02 +0000148 resetbuf:
Robert Griebl94a6a952002-05-19 19:00:14 +0000149 {
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000150 int i;
151 int e;
152 int o;
Robert Griebl94a6a952002-05-19 19:00:14 +0000153
Denis Vlasenko447b5432007-01-05 19:49:02 +0000154 o = posbits >> 3;
155 e = insize - o;
Robert Griebl94a6a952002-05-19 19:00:14 +0000156
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000157 for (i = 0; i < e; ++i)
158 inbuf[i] = inbuf[i + o];
Robert Griebl94a6a952002-05-19 19:00:14 +0000159
160 insize = e;
161 posbits = 0;
162 }
163
Bernhard Reutner-Fischer5d725462006-09-06 15:28:32 +0000164 if (insize < (int) (IBUFSIZ + 64) - IBUFSIZ) {
Glenn L McGrath1a2d75f2003-11-21 22:17:28 +0000165 rsize = safe_read(fd_in, inbuf + insize, IBUFSIZ);
Denis Vlasenko447b5432007-01-05 19:49:02 +0000166//error check??
Robert Griebl94a6a952002-05-19 19:00:14 +0000167 insize += rsize;
168 }
169
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000170 inbits = ((rsize > 0) ? (insize - insize % n_bits) << 3 :
171 (insize << 3) - (n_bits - 1));
Robert Griebl94a6a952002-05-19 19:00:14 +0000172
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000173 while (inbits > posbits) {
174 if (free_ent > maxcode) {
175 posbits =
176 ((posbits - 1) +
177 ((n_bits << 3) -
178 (posbits - 1 + (n_bits << 3)) % (n_bits << 3)));
Robert Griebl94a6a952002-05-19 19:00:14 +0000179 ++n_bits;
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000180 if (n_bits == maxbits) {
Robert Griebl94a6a952002-05-19 19:00:14 +0000181 maxcode = maxmaxcode;
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000182 } else {
183 maxcode = MAXCODE(n_bits) - 1;
184 }
185 bitmask = (1 << n_bits) - 1;
Robert Griebl94a6a952002-05-19 19:00:14 +0000186 goto resetbuf;
187 }
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000188 {
189 unsigned char *p = &inbuf[posbits >> 3];
Robert Griebl94a6a952002-05-19 19:00:14 +0000190
Denis Vlasenko447b5432007-01-05 19:49:02 +0000191 code = ((((long) (p[0])) | ((long) (p[1]) << 8) |
192 ((long) (p[2]) << 16)) >> (posbits & 0x7)) & bitmask;
Robert Griebl94a6a952002-05-19 19:00:14 +0000193 }
194 posbits += n_bits;
Robert Griebl94a6a952002-05-19 19:00:14 +0000195
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000196
197 if (oldcode == -1) {
Denis Vlasenko97a8dd32006-10-01 15:55:11 +0000198 oldcode = code;
199 finchar = (int) oldcode;
200 outbuf[outpos++] = (unsigned char) finchar;
Robert Griebl94a6a952002-05-19 19:00:14 +0000201 continue;
202 }
203
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000204 if (code == CLEAR && block_mode) {
Robert Griebl94a6a952002-05-19 19:00:14 +0000205 clear_tab_prefixof();
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000206 free_ent = FIRST - 1;
207 posbits =
208 ((posbits - 1) +
209 ((n_bits << 3) -
210 (posbits - 1 + (n_bits << 3)) % (n_bits << 3)));
Denis Vlasenko447b5432007-01-05 19:49:02 +0000211 n_bits = INIT_BITS;
212 maxcode = MAXCODE(INIT_BITS) - 1;
213 bitmask = (1 << INIT_BITS) - 1;
Robert Griebl94a6a952002-05-19 19:00:14 +0000214 goto resetbuf;
215 }
216
217 incode = code;
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000218 stackp = de_stack;
Robert Griebl94a6a952002-05-19 19:00:14 +0000219
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000220 /* Special case for KwKwK string. */
221 if (code >= free_ent) {
222 if (code > free_ent) {
223 unsigned char *p;
Robert Griebl94a6a952002-05-19 19:00:14 +0000224
225 posbits -= n_bits;
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000226 p = &inbuf[posbits >> 3];
Robert Griebl94a6a952002-05-19 19:00:14 +0000227
Manuel Novoa III cad53642003-03-19 09:13:01 +0000228 bb_error_msg
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000229 ("insize:%d posbits:%d inbuf:%02X %02X %02X %02X %02X (%d)",
230 insize, posbits, p[-1], p[0], p[1], p[2], p[3],
231 (posbits & 07));
Denys Vlasenkoccb88a62010-05-27 02:22:54 +0200232 bb_error_msg("corrupted data");
Denis Vlasenkoe8a07882007-06-10 15:08:44 +0000233 goto err;
Robert Griebl94a6a952002-05-19 19:00:14 +0000234 }
235
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000236 *--stackp = (unsigned char) finchar;
237 code = oldcode;
Robert Griebl94a6a952002-05-19 19:00:14 +0000238 }
239
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000240 /* Generate output characters in reverse order */
Denis Vlasenko87468852007-04-13 23:22:00 +0000241 while ((long) code >= (long) 256) {
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000242 *--stackp = tab_suffixof(code);
243 code = tab_prefixof(code);
Robert Griebl94a6a952002-05-19 19:00:14 +0000244 }
245
Denis Vlasenko447b5432007-01-05 19:49:02 +0000246 finchar = tab_suffixof(code);
247 *--stackp = (unsigned char) finchar;
Robert Griebl94a6a952002-05-19 19:00:14 +0000248
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000249 /* And put them out in forward order */
Robert Griebl94a6a952002-05-19 19:00:14 +0000250 {
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000251 int i;
Robert Griebl94a6a952002-05-19 19:00:14 +0000252
Denis Vlasenko447b5432007-01-05 19:49:02 +0000253 i = de_stack - stackp;
254 if (outpos + i >= OBUFSIZ) {
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000255 do {
256 if (i > OBUFSIZ - outpos) {
257 i = OBUFSIZ - outpos;
258 }
Robert Griebl94a6a952002-05-19 19:00:14 +0000259
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000260 if (i > 0) {
261 memcpy(outbuf + outpos, stackp, i);
Robert Griebl94a6a952002-05-19 19:00:14 +0000262 outpos += i;
263 }
264
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000265 if (outpos >= OBUFSIZ) {
Denis Vlasenko447b5432007-01-05 19:49:02 +0000266 full_write(fd_out, outbuf, outpos);
267//error check??
Denis Vlasenko5e34ff22009-04-21 11:09:40 +0000268 IF_DESKTOP(total_written += outpos;)
Robert Griebl94a6a952002-05-19 19:00:14 +0000269 outpos = 0;
270 }
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000271 stackp += i;
Denis Vlasenko447b5432007-01-05 19:49:02 +0000272 i = de_stack - stackp;
273 } while (i > 0);
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000274 } else {
275 memcpy(outbuf + outpos, stackp, i);
Robert Griebl94a6a952002-05-19 19:00:14 +0000276 outpos += i;
277 }
278 }
279
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000280 /* Generate the new entry. */
Denis Vlasenko447b5432007-01-05 19:49:02 +0000281 code = free_ent;
282 if (code < maxmaxcode) {
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000283 tab_prefixof(code) = (unsigned short) oldcode;
284 tab_suffixof(code) = (unsigned char) finchar;
285 free_ent = code + 1;
286 }
Robert Griebl94a6a952002-05-19 19:00:14 +0000287
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000288 /* Remember previous code. */
289 oldcode = incode;
Robert Griebl94a6a952002-05-19 19:00:14 +0000290 }
291
Glenn L McGrathfedbfe42002-11-28 09:09:47 +0000292 } while (rsize > 0);
Robert Griebl94a6a952002-05-19 19:00:14 +0000293
Glenn L McGrath2e41d0c2002-09-27 06:46:02 +0000294 if (outpos > 0) {
Denis Vlasenko447b5432007-01-05 19:49:02 +0000295 full_write(fd_out, outbuf, outpos);
296//error check??
Denis Vlasenko5e34ff22009-04-21 11:09:40 +0000297 IF_DESKTOP(total_written += outpos;)
Glenn L McGrath2e41d0c2002-09-27 06:46:02 +0000298 }
Robert Griebl94a6a952002-05-19 19:00:14 +0000299
Denis Vlasenko5e34ff22009-04-21 11:09:40 +0000300 retval = IF_DESKTOP(total_written) + 0;
Denis Vlasenkoe8a07882007-06-10 15:08:44 +0000301 err:
302 free(inbuf);
303 free(outbuf);
304 free(htab);
305 free(codetab);
306 return retval;
Robert Griebl94a6a952002-05-19 19:00:14 +0000307}