avoid silent data loss e.g., on NFS, due to unchecked close of stdout
[gzip.git] / unlzw.c
blob207023a151c970454aa0b7a5304d8f0abe3fef30
1 /* unlzw.c -- decompress files in LZW format.
2 * The code in this file is directly derived from the public domain 'compress'
3 * written by Spencer Thomas, Joe Orost, James Woods, Jim McKie, Steve Davies,
4 * Ken Turkowski, Dave Mack and Peter Jannesen.
6 * This is a temporary version which will be rewritten in some future version
7 * to accommodate in-memory decompression.
8 */
10 #ifdef RCSID
11 static char rcsid[] = "$Id$";
12 #endif
14 #include <config.h>
15 #include "tailor.h"
17 #ifdef HAVE_UNISTD_H
18 # include <unistd.h>
19 #endif
20 #ifdef HAVE_FCNTL_H
21 # include <fcntl.h>
22 #endif
24 #include "gzip.h"
25 #include "lzw.h"
27 typedef unsigned char char_type;
28 typedef long code_int;
29 typedef unsigned long count_int;
30 typedef unsigned short count_short;
31 typedef unsigned long cmp_code_int;
33 #define MAXCODE(n) (1L << (n))
35 #ifndef REGISTERS
36 # define REGISTERS 2
37 #endif
38 #define REG1
39 #define REG2
40 #define REG3
41 #define REG4
42 #define REG5
43 #define REG6
44 #define REG7
45 #define REG8
46 #define REG9
47 #define REG10
48 #define REG11
49 #define REG12
50 #define REG13
51 #define REG14
52 #define REG15
53 #define REG16
54 #if REGISTERS >= 1
55 # undef REG1
56 # define REG1 register
57 #endif
58 #if REGISTERS >= 2
59 # undef REG2
60 # define REG2 register
61 #endif
62 #if REGISTERS >= 3
63 # undef REG3
64 # define REG3 register
65 #endif
66 #if REGISTERS >= 4
67 # undef REG4
68 # define REG4 register
69 #endif
70 #if REGISTERS >= 5
71 # undef REG5
72 # define REG5 register
73 #endif
74 #if REGISTERS >= 6
75 # undef REG6
76 # define REG6 register
77 #endif
78 #if REGISTERS >= 7
79 # undef REG7
80 # define REG7 register
81 #endif
82 #if REGISTERS >= 8
83 # undef REG8
84 # define REG8 register
85 #endif
86 #if REGISTERS >= 9
87 # undef REG9
88 # define REG9 register
89 #endif
90 #if REGISTERS >= 10
91 # undef REG10
92 # define REG10 register
93 #endif
94 #if REGISTERS >= 11
95 # undef REG11
96 # define REG11 register
97 #endif
98 #if REGISTERS >= 12
99 # undef REG12
100 # define REG12 register
101 #endif
102 #if REGISTERS >= 13
103 # undef REG13
104 # define REG13 register
105 #endif
106 #if REGISTERS >= 14
107 # undef REG14
108 # define REG14 register
109 #endif
110 #if REGISTERS >= 15
111 # undef REG15
112 # define REG15 register
113 #endif
114 #if REGISTERS >= 16
115 # undef REG16
116 # define REG16 register
117 #endif
119 #ifndef BYTEORDER
120 # define BYTEORDER 0000
121 #endif
123 #ifndef NOALLIGN
124 # define NOALLIGN 0
125 #endif
128 union bytes {
129 long word;
130 struct {
131 #if BYTEORDER == 4321
132 char_type b1;
133 char_type b2;
134 char_type b3;
135 char_type b4;
136 #else
137 #if BYTEORDER == 1234
138 char_type b4;
139 char_type b3;
140 char_type b2;
141 char_type b1;
142 #else
143 # undef BYTEORDER
144 int dummy;
145 #endif
146 #endif
147 } bytes;
150 #if BYTEORDER == 4321 && NOALLIGN == 1
151 # define input(b,o,c,n,m){ \
152 (c) = (*(long *)(&(b)[(o)>>3])>>((o)&0x7))&(m); \
153 (o) += (n); \
155 #else
156 # define input(b,o,c,n,m){ \
157 REG1 char_type *p = &(b)[(o)>>3]; \
158 (c) = ((((long)(p[0]))|((long)(p[1])<<8)| \
159 ((long)(p[2])<<16))>>((o)&0x7))&(m); \
160 (o) += (n); \
162 #endif
164 #ifndef MAXSEG_64K
165 /* DECLARE(ush, tab_prefix, (1<<BITS)); -- prefix code */
166 # define tab_prefixof(i) tab_prefix[i]
167 # define clear_tab_prefixof() memzero(tab_prefix, 256);
168 #else
169 /* DECLARE(ush, tab_prefix0, (1<<(BITS-1)); -- prefix for even codes */
170 /* DECLARE(ush, tab_prefix1, (1<<(BITS-1)); -- prefix for odd codes */
171 ush *tab_prefix[2];
172 # define tab_prefixof(i) tab_prefix[(i)&1][(i)>>1]
173 # define clear_tab_prefixof() \
174 memzero(tab_prefix0, 128), \
175 memzero(tab_prefix1, 128);
176 #endif
177 #define de_stack ((char_type *)(&d_buf[DIST_BUFSIZE-1]))
178 #define tab_suffixof(i) tab_suffix[i]
180 int block_mode = BLOCK_MODE; /* block compress mode -C compatible with 2.0 */
182 /* ============================================================================
183 * Decompress in to out. This routine adapts to the codes in the
184 * file building the "string" table on-the-fly; requiring no table to
185 * be stored in the compressed file.
186 * IN assertions: the buffer inbuf contains already the beginning of
187 * the compressed data, from offsets iptr to insize-1 included.
188 * The magic header has already been checked and skipped.
189 * bytes_in and bytes_out have been initialized.
191 int unlzw(in, out)
192 int in, out; /* input and output file descriptors */
194 REG2 char_type *stackp;
195 REG3 code_int code;
196 REG4 int finchar;
197 REG5 code_int oldcode;
198 REG6 code_int incode;
199 REG7 long inbits;
200 REG8 long posbits;
201 REG9 int outpos;
202 /* REG10 int insize; (global) */
203 REG11 unsigned bitmask;
204 REG12 code_int free_ent;
205 REG13 code_int maxcode;
206 REG14 code_int maxmaxcode;
207 REG15 int n_bits;
208 REG16 int rsize;
210 #ifdef MAXSEG_64K
211 tab_prefix[0] = tab_prefix0;
212 tab_prefix[1] = tab_prefix1;
213 #endif
214 maxbits = get_byte();
215 block_mode = maxbits & BLOCK_MODE;
216 if ((maxbits & LZW_RESERVED) != 0) {
217 WARN((stderr, "\n%s: %s: warning, unknown flags 0x%x\n",
218 program_name, ifname, maxbits & LZW_RESERVED));
220 maxbits &= BIT_MASK;
221 maxmaxcode = MAXCODE(maxbits);
223 if (maxbits > BITS) {
224 fprintf(stderr,
225 "\n%s: %s: compressed with %d bits, can only handle %d bits\n",
226 program_name, ifname, maxbits, BITS);
227 exit_code = ERROR;
228 return ERROR;
230 rsize = insize;
231 maxcode = MAXCODE(n_bits = INIT_BITS)-1;
232 bitmask = (1<<n_bits)-1;
233 oldcode = -1;
234 finchar = 0;
235 outpos = 0;
236 posbits = inptr<<3;
238 free_ent = ((block_mode) ? FIRST : 256);
240 clear_tab_prefixof(); /* Initialize the first 256 entries in the table. */
242 for (code = 255 ; code >= 0 ; --code) {
243 tab_suffixof(code) = (char_type)code;
245 do {
246 REG1 int i;
247 int e;
248 int o;
250 resetbuf:
251 e = insize-(o = (posbits>>3));
253 for (i = 0 ; i < e ; ++i) {
254 inbuf[i] = inbuf[i+o];
256 insize = e;
257 posbits = 0;
259 if (insize < INBUF_EXTRA) {
260 rsize = read_buffer (in, (char *) inbuf + insize, INBUFSIZ);
261 if (rsize == -1) {
262 read_error();
264 insize += rsize;
265 bytes_in += (off_t)rsize;
267 inbits = ((rsize != 0) ? ((long)insize - insize%n_bits)<<3 :
268 ((long)insize<<3)-(n_bits-1));
270 while (inbits > posbits) {
271 if (free_ent > maxcode) {
272 posbits = ((posbits-1) +
273 ((n_bits<<3)-(posbits-1+(n_bits<<3))%(n_bits<<3)));
274 ++n_bits;
275 if (n_bits == maxbits) {
276 maxcode = maxmaxcode;
277 } else {
278 maxcode = MAXCODE(n_bits)-1;
280 bitmask = (1<<n_bits)-1;
281 goto resetbuf;
283 input(inbuf,posbits,code,n_bits,bitmask);
284 Tracev((stderr, "%d ", code));
286 if (oldcode == -1) {
287 if (256 <= code)
288 gzip_error ("corrupt input.");
289 outbuf[outpos++] = (char_type)(finchar = (int)(oldcode=code));
290 continue;
292 if (code == CLEAR && block_mode) {
293 clear_tab_prefixof();
294 free_ent = FIRST - 1;
295 posbits = ((posbits-1) +
296 ((n_bits<<3)-(posbits-1+(n_bits<<3))%(n_bits<<3)));
297 maxcode = MAXCODE(n_bits = INIT_BITS)-1;
298 bitmask = (1<<n_bits)-1;
299 goto resetbuf;
301 incode = code;
302 stackp = de_stack;
304 if (code >= free_ent) { /* Special case for KwKwK string. */
305 if (code > free_ent) {
306 #ifdef DEBUG
307 char_type *p;
309 posbits -= n_bits;
310 p = &inbuf[posbits>>3];
311 fprintf(stderr,
312 "code:%ld free_ent:%ld n_bits:%d insize:%u\n",
313 code, free_ent, n_bits, insize);
314 fprintf(stderr,
315 "posbits:%ld inbuf:%02X %02X %02X %02X %02X\n",
316 posbits, p[-1],p[0],p[1],p[2],p[3]);
317 #endif
318 if (!test && outpos > 0) {
319 write_buf(out, (char*)outbuf, outpos);
320 bytes_out += (off_t)outpos;
322 gzip_error (to_stdout
323 ? "corrupt input."
324 : "corrupt input. Use zcat to recover some data.");
326 *--stackp = (char_type)finchar;
327 code = oldcode;
330 while ((cmp_code_int)code >= (cmp_code_int)256) {
331 /* Generate output characters in reverse order */
332 *--stackp = tab_suffixof(code);
333 code = tab_prefixof(code);
335 *--stackp = (char_type)(finchar = tab_suffixof(code));
337 /* And put them out in forward order */
339 REG1 int i;
341 if (outpos+(i = (de_stack-stackp)) >= OUTBUFSIZ) {
342 do {
343 if (i > OUTBUFSIZ-outpos) i = OUTBUFSIZ-outpos;
345 if (i > 0) {
346 memcpy(outbuf+outpos, stackp, i);
347 outpos += i;
349 if (outpos >= OUTBUFSIZ) {
350 if (!test) {
351 write_buf(out, (char*)outbuf, outpos);
352 bytes_out += (off_t)outpos;
354 outpos = 0;
356 stackp+= i;
357 } while ((i = (de_stack-stackp)) > 0);
358 } else {
359 memcpy(outbuf+outpos, stackp, i);
360 outpos += i;
364 if ((code = free_ent) < maxmaxcode) { /* Generate the new entry. */
366 tab_prefixof(code) = (unsigned short)oldcode;
367 tab_suffixof(code) = (char_type)finchar;
368 free_ent = code+1;
370 oldcode = incode; /* Remember previous code. */
372 } while (rsize != 0);
374 if (!test && outpos > 0) {
375 write_buf(out, (char*)outbuf, outpos);
376 bytes_out += (off_t)outpos;
378 return OK;