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.
19 typedef unsigned char char_type
;
20 typedef long code_int
;
21 typedef unsigned long count_int
;
22 typedef unsigned short count_short
;
23 typedef unsigned long cmp_code_int
;
25 #define MAXCODE(n) (1L << (n))
28 # define BYTEORDER 0000
58 #if BYTEORDER == 4321 && NOALLIGN == 1
59 # define input(b,o,c,n,m){ \
60 (c) = (*(long *)(&(b)[(o)>>3])>>((o)&0x7))&(m); \
64 # define input(b,o,c,n,m){ \
65 char_type *p = &(b)[(o)>>3]; \
66 (c) = ((((long)(p[0]))|((long)(p[1])<<8)| \
67 ((long)(p[2])<<16))>>((o)&0x7))&(m); \
73 /* DECLARE(ush, tab_prefix, (1<<BITS)); -- prefix code */
74 # define tab_prefixof(i) tab_prefix[i]
75 # define clear_tab_prefixof() memzero(tab_prefix, 256);
77 /* DECLARE(ush, tab_prefix0, (1<<(BITS-1)); -- prefix for even codes */
78 /* DECLARE(ush, tab_prefix1, (1<<(BITS-1)); -- prefix for odd codes */
80 # define tab_prefixof(i) tab_prefix[(i)&1][(i)>>1]
81 # define clear_tab_prefixof() \
82 memzero(tab_prefix0, 128), \
83 memzero(tab_prefix1, 128);
85 #define de_stack ((char_type *)(&d_buf[DIST_BUFSIZE-1]))
86 #define tab_suffixof(i) tab_suffix[i]
88 /* block compress mode -C compatible with 2.0 */
89 static int block_mode
= BLOCK_MODE
;
91 /* ============================================================================
92 * Decompress in to out. This routine adapts to the codes in the
93 * file building the "string" table on-the-fly; requiring no table to
94 * be stored in the compressed file.
95 * IN assertions: the buffer inbuf contains already the beginning of
96 * the compressed data, from offsets iptr to insize-1 included.
97 * The magic header has already been checked and skipped.
98 * bytes_in and bytes_out have been initialized.
101 int in
, out
; /* input and output file descriptors */
111 /* int insize; (global) */
120 tab_prefix
[0] = tab_prefix0
;
121 tab_prefix
[1] = tab_prefix1
;
123 maxbits
= get_byte();
124 block_mode
= maxbits
& BLOCK_MODE
;
125 if ((maxbits
& LZW_RESERVED
) != 0) {
126 WARN((stderr
, "\n%s: %s: warning, unknown flags 0x%x\n",
127 program_name
, ifname
, (unsigned int) maxbits
& LZW_RESERVED
));
130 maxmaxcode
= MAXCODE(maxbits
);
132 if (maxbits
> BITS
) {
134 "\n%s: %s: compressed with %d bits, can only handle %d bits\n",
135 program_name
, ifname
, maxbits
, BITS
);
140 maxcode
= MAXCODE(n_bits
= INIT_BITS
)-1;
141 bitmask
= (1<<n_bits
)-1;
147 free_ent
= ((block_mode
) ? FIRST
: 256);
149 clear_tab_prefixof(); /* Initialize the first 256 entries in the table. */
151 for (code
= 255 ; code
>= 0 ; --code
) {
152 tab_suffixof(code
) = (char_type
)code
;
161 e
= o
<= insize
? insize
- o
: 0;
163 for (i
= 0 ; i
< e
; ++i
) {
164 inbuf
[i
] = inbuf
[i
+o
];
169 if (insize
< INBUF_EXTRA
) {
170 rsize
= read_buffer (in
, (char *) inbuf
+ insize
, INBUFSIZ
);
175 bytes_in
+= (off_t
)rsize
;
177 inbits
= ((rsize
!= 0) ? ((long)insize
- insize
%n_bits
)<<3 :
178 ((long)insize
<<3)-(n_bits
-1));
180 while (inbits
> posbits
) {
181 if (free_ent
> maxcode
) {
182 posbits
= ((posbits
-1) +
183 ((n_bits
<<3)-(posbits
-1+(n_bits
<<3))%(n_bits
<<3)));
185 if (n_bits
== maxbits
) {
186 maxcode
= maxmaxcode
;
188 maxcode
= MAXCODE(n_bits
)-1;
190 bitmask
= (1<<n_bits
)-1;
193 input(inbuf
,posbits
,code
,n_bits
,bitmask
);
194 Tracev((stderr
, "%ld ", code
));
198 gzip_error ("corrupt input.");
199 outbuf
[outpos
++] = (char_type
)(finchar
= (int)(oldcode
=code
));
202 if (code
== CLEAR
&& block_mode
) {
203 clear_tab_prefixof();
204 free_ent
= FIRST
- 1;
205 posbits
= ((posbits
-1) +
206 ((n_bits
<<3)-(posbits
-1+(n_bits
<<3))%(n_bits
<<3)));
207 maxcode
= MAXCODE(n_bits
= INIT_BITS
)-1;
208 bitmask
= (1<<n_bits
)-1;
214 if (code
>= free_ent
) { /* Special case for KwKwK string. */
215 if (code
> free_ent
) {
220 p
= &inbuf
[posbits
>>3];
222 "code:%ld free_ent:%ld n_bits:%d insize:%u\n",
223 code
, free_ent
, n_bits
, insize
);
225 "posbits:%ld inbuf:%02X %02X %02X %02X %02X\n",
226 posbits
, p
[-1],p
[0],p
[1],p
[2],p
[3]);
229 write_buf (out
, outbuf
, outpos
);
230 gzip_error (to_stdout
232 : "corrupt input. Use zcat to recover some data.");
234 *--stackp
= (char_type
)finchar
;
238 while ((cmp_code_int
)code
>= (cmp_code_int
)256) {
239 /* Generate output characters in reverse order */
240 *--stackp
= tab_suffixof(code
);
241 code
= tab_prefixof(code
);
243 *--stackp
= (char_type
)(finchar
= tab_suffixof(code
));
245 /* And put them out in forward order */
249 if (outpos
+(i
= (de_stack
-stackp
)) >= OUTBUFSIZ
) {
251 if (i
> OUTBUFSIZ
-outpos
) i
= OUTBUFSIZ
-outpos
;
254 memcpy(outbuf
+outpos
, stackp
, i
);
257 if (outpos
>= OUTBUFSIZ
) {
258 write_buf (out
, outbuf
, outpos
);
262 } while ((i
= (de_stack
-stackp
)) > 0);
264 memcpy(outbuf
+outpos
, stackp
, i
);
269 if ((code
= free_ent
) < maxmaxcode
) { /* Generate the new entry. */
271 tab_prefixof(code
) = (unsigned short)oldcode
;
272 tab_suffixof(code
) = (char_type
)finchar
;
275 oldcode
= incode
; /* Remember previous code. */
277 } while (rsize
!= 0);
280 write_buf (out
, outbuf
, outpos
);