1 |
90 |
jeremybenn |
/* mycompress.c. Test of Or1ksim compression program
|
2 |
|
|
|
3 |
|
|
Copyright (C) 1999-2006 OpenCores
|
4 |
|
|
Copyright (C) 2010 Embecosm Limited
|
5 |
|
|
|
6 |
|
|
Contributors various OpenCores participants
|
7 |
|
|
Contributor Jeremy Bennett <jeremy.bennett@embecosm.com>
|
8 |
|
|
|
9 |
|
|
This file is part of OpenRISC 1000 Architectural Simulator.
|
10 |
|
|
|
11 |
|
|
This program is free software; you can redistribute it and/or modify it
|
12 |
|
|
under the terms of the GNU General Public License as published by the Free
|
13 |
|
|
Software Foundation; either version 3 of the License, or (at your option)
|
14 |
|
|
any later version.
|
15 |
|
|
|
16 |
|
|
This program is distributed in the hope that it will be useful, but WITHOUT
|
17 |
|
|
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
|
18 |
|
|
FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
|
19 |
|
|
more details.
|
20 |
|
|
|
21 |
|
|
You should have received a copy of the GNU General Public License along
|
22 |
|
|
with this program. If not, see <http: www.gnu.org/licenses/>. */
|
23 |
|
|
|
24 |
|
|
/* ----------------------------------------------------------------------------
|
25 |
|
|
This code is commented throughout for use with Doxygen.
|
26 |
|
|
--------------------------------------------------------------------------*/
|
27 |
|
|
|
28 |
|
|
#include <stdarg.h>
|
29 |
|
|
#include "support.h"
|
30 |
|
|
|
31 |
|
|
#define BYTES_TO_COMPRESS 1000
|
32 |
|
|
|
33 |
|
|
/*
|
34 |
|
|
* Compress - data compression program
|
35 |
|
|
*/
|
36 |
|
|
#define min(a,b) ((a>b) ? b : a)
|
37 |
|
|
|
38 |
|
|
/*
|
39 |
|
|
* machine variants which require cc -Dmachine: pdp11, z8000, pcxt
|
40 |
|
|
*/
|
41 |
|
|
|
42 |
|
|
/*
|
43 |
|
|
* Set USERMEM to the maximum amount of physical user memory available
|
44 |
|
|
* in bytes. USERMEM is used to determine the maximum BITS that can be used
|
45 |
|
|
* for compression.
|
46 |
|
|
*
|
47 |
|
|
* SACREDMEM is the amount of physical memory saved for others; compress
|
48 |
|
|
* will hog the rest.
|
49 |
|
|
*/
|
50 |
|
|
#ifndef SACREDMEM
|
51 |
|
|
#define SACREDMEM 0
|
52 |
|
|
#endif
|
53 |
|
|
|
54 |
|
|
/* #ifndef USERMEM */
|
55 |
|
|
#define USERMEM 60000 /* default user memory */
|
56 |
|
|
/* #endif */
|
57 |
|
|
|
58 |
|
|
#ifdef interdata /* (Perkin-Elmer) */
|
59 |
|
|
#define SIGNED_COMPARE_SLOW /* signed compare is slower than unsigned */
|
60 |
|
|
#endif
|
61 |
|
|
|
62 |
|
|
#ifdef USERMEM
|
63 |
|
|
# if USERMEM >= (433484+SACREDMEM)
|
64 |
|
|
# define PBITS 16
|
65 |
|
|
# else
|
66 |
|
|
# if USERMEM >= (229600+SACREDMEM)
|
67 |
|
|
# define PBITS 15
|
68 |
|
|
# else
|
69 |
|
|
# if USERMEM >= (127536+SACREDMEM)
|
70 |
|
|
# define PBITS 14
|
71 |
|
|
# else
|
72 |
|
|
# if USERMEM >= (73464+SACREDMEM)
|
73 |
|
|
# define PBITS 13
|
74 |
|
|
# else
|
75 |
|
|
# define PBITS 12
|
76 |
|
|
# endif
|
77 |
|
|
# endif
|
78 |
|
|
# endif
|
79 |
|
|
# endif
|
80 |
|
|
# undef USERMEM
|
81 |
|
|
#endif /* USERMEM */
|
82 |
|
|
|
83 |
|
|
#ifdef PBITS /* Preferred BITS for this memory size */
|
84 |
|
|
# ifndef BITS
|
85 |
|
|
# define BITS PBITS
|
86 |
|
|
# endif /* BITS */
|
87 |
|
|
#endif /* PBITS */
|
88 |
|
|
|
89 |
|
|
#if BITS == 16
|
90 |
|
|
# define HSIZE 69001 /* 95% occupancy */
|
91 |
|
|
#endif
|
92 |
|
|
#if BITS == 15
|
93 |
|
|
# define HSIZE 35023 /* 94% occupancy */
|
94 |
|
|
#endif
|
95 |
|
|
#if BITS == 14
|
96 |
|
|
# define HSIZE 18013 /* 91% occupancy */
|
97 |
|
|
#endif
|
98 |
|
|
#if BITS == 13
|
99 |
|
|
# define HSIZE 9001 /* 91% occupancy */
|
100 |
|
|
#endif
|
101 |
|
|
#if BITS <= 12
|
102 |
|
|
# define HSIZE 5003 /* 80% occupancy */
|
103 |
|
|
#endif
|
104 |
|
|
|
105 |
|
|
/*
|
106 |
|
|
* a code_int must be able to hold 2**BITS values of type int, and also -1
|
107 |
|
|
*/
|
108 |
|
|
#if BITS > 15
|
109 |
|
|
typedef long int code_int;
|
110 |
|
|
#else
|
111 |
|
|
typedef int code_int;
|
112 |
|
|
#endif
|
113 |
|
|
|
114 |
|
|
#ifdef SIGNED_COMPARE_SLOW
|
115 |
|
|
typedef unsigned long int count_int;
|
116 |
|
|
typedef unsigned short int count_short;
|
117 |
|
|
#else
|
118 |
|
|
typedef long int count_int;
|
119 |
|
|
#endif
|
120 |
|
|
|
121 |
|
|
#ifdef NO_UCHAR
|
122 |
|
|
typedef char char_type;
|
123 |
|
|
#else
|
124 |
|
|
typedef unsigned char char_type;
|
125 |
|
|
#endif /* UCHAR */
|
126 |
|
|
char_type magic_header[] = { "\037\235" }; /* 1F 9D */
|
127 |
|
|
|
128 |
|
|
/* Defines for third byte of header */
|
129 |
|
|
#define BIT_MASK 0x1f
|
130 |
|
|
#define BLOCK_MASK 0x80
|
131 |
|
|
/* Masks 0x40 and 0x20 are free. I think 0x20 should mean that there is
|
132 |
|
|
a fourth header byte (for expansion).
|
133 |
|
|
*/
|
134 |
|
|
#define INIT_BITS 9 /* initial number of bits/code */
|
135 |
|
|
|
136 |
|
|
/*
|
137 |
|
|
* compress.c - File compression ala IEEE Computer, June 1984.
|
138 |
|
|
*
|
139 |
|
|
* Authors: Spencer W. Thomas (decvax!harpo!utah-cs!utah-gr!thomas)
|
140 |
|
|
* Jim McKie (decvax!mcvax!jim)
|
141 |
|
|
* Steve Davies (decvax!vax135!petsd!peora!srd)
|
142 |
|
|
* Ken Turkowski (decvax!decwrl!turtlevax!ken)
|
143 |
|
|
* James A. Woods (decvax!ihnp4!ames!jaw)
|
144 |
|
|
* Joe Orost (decvax!vax135!petsd!joe)
|
145 |
|
|
*
|
146 |
|
|
*/
|
147 |
|
|
|
148 |
|
|
#if i386
|
149 |
|
|
#include <stdio.h>
|
150 |
|
|
#include <stdio.h>
|
151 |
|
|
#include <ctype.h>
|
152 |
|
|
#include <signal.h>
|
153 |
|
|
#include <sys/types.h>
|
154 |
|
|
#include <sys/stat.h>
|
155 |
|
|
#endif
|
156 |
|
|
|
157 |
|
|
#define ARGVAL() (*++(*argv) || (--argc && *++argv))
|
158 |
|
|
|
159 |
|
|
int n_bits; /* number of bits/code */
|
160 |
|
|
int maxbits = BITS; /* user settable max # bits/code */
|
161 |
|
|
code_int maxcode; /* maximum code, given n_bits */
|
162 |
|
|
code_int maxmaxcode = 1L << BITS; /* should NEVER generate this code */
|
163 |
|
|
#ifdef COMPATIBLE /* But wrong! */
|
164 |
|
|
# define MAXCODE(n_bits) (1L << (n_bits) - 1)
|
165 |
|
|
#else
|
166 |
|
|
# define MAXCODE(n_bits) ((1L << (n_bits)) - 1)
|
167 |
|
|
#endif /* COMPATIBLE */
|
168 |
|
|
|
169 |
|
|
# ifdef sel
|
170 |
|
|
/* support gould base register problems */
|
171 |
|
|
/*NOBASE*/
|
172 |
|
|
count_int htab [HSIZE];
|
173 |
|
|
unsigned short codetab [HSIZE];
|
174 |
|
|
/*NOBASE*/
|
175 |
|
|
# else /* !gould */
|
176 |
|
|
count_int htab [HSIZE];
|
177 |
|
|
unsigned short codetab [HSIZE];
|
178 |
|
|
# endif /* !gould */
|
179 |
|
|
#define htabof(i) htab[i]
|
180 |
|
|
#define codetabof(i) codetab[i]
|
181 |
|
|
code_int hsize = HSIZE; /* for dynamic table sizing */
|
182 |
|
|
count_int fsize;
|
183 |
|
|
|
184 |
|
|
/*
|
185 |
|
|
* To save much memory, we overlay the table used by compress() with those
|
186 |
|
|
* used by decompress(). The tab_prefix table is the same size and type
|
187 |
|
|
* as the codetab. The tab_suffix table needs 2**BITS characters. We
|
188 |
|
|
* get this from the beginning of htab. The output stack uses the rest
|
189 |
|
|
* of htab, and contains characters. There is plenty of room for any
|
190 |
|
|
* possible stack (stack used to be 8000 characters).
|
191 |
|
|
*/
|
192 |
|
|
|
193 |
|
|
#define tab_prefixof(i) codetabof(i)
|
194 |
|
|
#ifdef XENIX_16
|
195 |
|
|
# define tab_suffixof(i) ((char_type *)htab[(i)>>15])[(i) & 0x7fff]
|
196 |
|
|
# define de_stack ((char_type *)(htab2))
|
197 |
|
|
#else /* Normal machine */
|
198 |
|
|
# define tab_suffixof(i) ((char_type *)(htab))[i]
|
199 |
|
|
# define de_stack ((char_type *)&tab_suffixof(1<<BITS))
|
200 |
|
|
#endif /* XENIX_16 */
|
201 |
|
|
|
202 |
|
|
code_int free_ent = 0; /* first unused entry */
|
203 |
|
|
int exit_stat = 0;
|
204 |
|
|
|
205 |
|
|
code_int getcode();
|
206 |
|
|
|
207 |
|
|
int nomagic = 0; /* Use a 3-byte magic number header, unless old file */
|
208 |
|
|
int zcat_flg = 0; /* Write output on stdout, suppress messages */
|
209 |
|
|
int quiet = 1; /* don't tell me about compression */
|
210 |
|
|
|
211 |
|
|
/*
|
212 |
|
|
* block compression parameters -- after all codes are used up,
|
213 |
|
|
* and compression rate changes, start over.
|
214 |
|
|
*/
|
215 |
|
|
int block_compress = BLOCK_MASK;
|
216 |
|
|
int clear_flg = 0;
|
217 |
|
|
long int ratio = 0;
|
218 |
|
|
#define CHECK_GAP 10000 /* ratio check interval */
|
219 |
|
|
count_int checkpoint = CHECK_GAP;
|
220 |
|
|
/*
|
221 |
|
|
* the next two codes should not be changed lightly, as they must not
|
222 |
|
|
* lie within the contiguous general code space.
|
223 |
|
|
*/
|
224 |
|
|
#define FIRST 257 /* first free entry */
|
225 |
|
|
#define CLEAR 256 /* table clear output code */
|
226 |
|
|
|
227 |
|
|
int force = 0;
|
228 |
|
|
char ofname [100];
|
229 |
|
|
int (*bgnd_flag)();
|
230 |
|
|
|
231 |
|
|
/* reset code table */
|
232 |
|
|
void cl_hash(register count_int hsize);
|
233 |
|
|
/* table clear for block compress */
|
234 |
|
|
void cl_block ();
|
235 |
|
|
void output(code_int code);
|
236 |
|
|
|
237 |
|
|
|
238 |
|
|
int do_decomp = 0;
|
239 |
|
|
|
240 |
|
|
static int offset;
|
241 |
|
|
long int in_count = 1; /* length of input */
|
242 |
|
|
long int bytes_out; /* length of compressed output */
|
243 |
|
|
long int out_count = 0; /* # of codes output (for debugging) */
|
244 |
|
|
|
245 |
|
|
/*
|
246 |
|
|
* compress stdin to stdout
|
247 |
|
|
*
|
248 |
|
|
* Algorithm: use open addressing double hashing (no chaining) on the
|
249 |
|
|
* prefix code / next character combination. We do a variant of Knuth's
|
250 |
|
|
* algorithm D (vol. 3, sec. 6.4) along with G. Knott's relatively-prime
|
251 |
|
|
* secondary probe. Here, the modular division first probe is gives way
|
252 |
|
|
* to a faster exclusive-or manipulation. Also do block compression with
|
253 |
|
|
* an adaptive reset, whereby the code table is cleared when the compression
|
254 |
|
|
* ratio decreases, but after the table fills. The variable-length output
|
255 |
|
|
* codes are re-sized at this point, and a special CLEAR code is generated
|
256 |
|
|
* for the decompressor. Late addition: construct the table according to
|
257 |
|
|
* file size for noticeable speed improvement on small files. Please direct
|
258 |
|
|
* questions about this implementation to ames!jaw.
|
259 |
|
|
*/
|
260 |
|
|
|
261 |
|
|
int main() {
|
262 |
|
|
register long fcode;
|
263 |
|
|
register code_int i = 0;
|
264 |
|
|
register int c;
|
265 |
|
|
register code_int ent;
|
266 |
|
|
#ifdef XENIX_16
|
267 |
|
|
register code_int disp;
|
268 |
|
|
#else /* Normal machine */
|
269 |
|
|
register int disp;
|
270 |
|
|
#endif
|
271 |
|
|
register code_int hsize_reg;
|
272 |
|
|
register int hshift;
|
273 |
|
|
|
274 |
|
|
#ifndef COMPATIBLE
|
275 |
|
|
if (nomagic == 0) {
|
276 |
|
|
/* putchar(magic_header[0]); putchar(magic_header[1]);
|
277 |
|
|
putchar((char)(maxbits | block_compress)); */
|
278 |
|
|
}
|
279 |
|
|
#endif /* COMPATIBLE */
|
280 |
|
|
|
281 |
|
|
offset = 0;
|
282 |
|
|
bytes_out = 3; /* includes 3-byte header mojo */
|
283 |
|
|
out_count = 0;
|
284 |
|
|
clear_flg = 0;
|
285 |
|
|
ratio = 0;
|
286 |
|
|
in_count = 1;
|
287 |
|
|
|
288 |
|
|
printf("main: bytes_out %d... hsize %d\n", (int)bytes_out, (int)hsize);
|
289 |
|
|
|
290 |
|
|
checkpoint = CHECK_GAP;
|
291 |
|
|
maxcode = MAXCODE(n_bits = INIT_BITS);
|
292 |
|
|
free_ent = ((block_compress) ? FIRST : 256 );
|
293 |
|
|
|
294 |
|
|
|
295 |
|
|
ent = '\0'; /* getchar (); */
|
296 |
|
|
|
297 |
|
|
hshift = 0;
|
298 |
|
|
for ( fcode = (long) hsize; fcode < 65536L; fcode *= 2L )
|
299 |
|
|
hshift++;
|
300 |
|
|
hshift = 8 - hshift; /* set hash code range bound */
|
301 |
|
|
printf("main: hshift %d...\n", hshift);
|
302 |
|
|
|
303 |
|
|
hsize_reg = hsize;
|
304 |
|
|
cl_hash( (count_int) hsize_reg); /* clear hash table */
|
305 |
|
|
|
306 |
|
|
/*#ifdef SIGNED_COMPARE_SLOW
|
307 |
|
|
while ( (c = getchar()) != (unsigned) EOF ) {
|
308 |
|
|
#else
|
309 |
|
|
while ( (c = getchar()) != EOF ) {
|
310 |
|
|
#endif*/
|
311 |
|
|
printf("main: bytes_out %d...\n", (int)bytes_out);
|
312 |
|
|
printf("main: hsize_reg %d...\n", (int)hsize_reg);
|
313 |
|
|
printf("main: before compress %d...\n", (int)in_count);
|
314 |
|
|
while (in_count < BYTES_TO_COMPRESS) {
|
315 |
|
|
c = in_count % 255;
|
316 |
|
|
|
317 |
|
|
printf("main: compressing %d...\n", (int)in_count);
|
318 |
|
|
in_count++;
|
319 |
|
|
fcode = (long) (((long) c << maxbits) + ent);
|
320 |
|
|
i = (((long)c << hshift) ^ ent); /* xor hashing */
|
321 |
|
|
|
322 |
|
|
if ( htabof (i) == fcode ) {
|
323 |
|
|
ent = codetabof (i);
|
324 |
|
|
continue;
|
325 |
|
|
} else if ( (long)htabof (i) < 0 ) /* empty slot */
|
326 |
|
|
goto nomatch;
|
327 |
|
|
|
328 |
|
|
disp = hsize_reg - i; /* secondary hash (after G. Knott) */
|
329 |
|
|
if ( i == 0 )
|
330 |
|
|
disp = 1;
|
331 |
|
|
probe:
|
332 |
|
|
if ( (i -= disp) < 0 )
|
333 |
|
|
i += hsize_reg;
|
334 |
|
|
|
335 |
|
|
if ( htabof (i) == fcode ) {
|
336 |
|
|
ent = codetabof (i);
|
337 |
|
|
continue;
|
338 |
|
|
}
|
339 |
|
|
if ( (long)htabof (i) > 0 )
|
340 |
|
|
goto probe;
|
341 |
|
|
nomatch:
|
342 |
|
|
output ( (code_int) ent );
|
343 |
|
|
out_count++;
|
344 |
|
|
ent = c;
|
345 |
|
|
#ifdef SIGNED_COMPARE_SLOW
|
346 |
|
|
if ( (unsigned) free_ent < (unsigned) maxmaxcode) {
|
347 |
|
|
#else
|
348 |
|
|
if ( free_ent < maxmaxcode ) {
|
349 |
|
|
#endif
|
350 |
|
|
codetabof (i) = free_ent++; /* code -> hashtable */
|
351 |
|
|
htabof (i) = fcode;
|
352 |
|
|
}
|
353 |
|
|
else if ( (count_int)in_count >= checkpoint && block_compress )
|
354 |
|
|
cl_block ();
|
355 |
|
|
}
|
356 |
|
|
/*
|
357 |
|
|
* Put out the final code.
|
358 |
|
|
*/
|
359 |
|
|
printf("main: output...\n");
|
360 |
|
|
output( (code_int)ent );
|
361 |
|
|
out_count++;
|
362 |
|
|
output( (code_int)-1 );
|
363 |
|
|
|
364 |
|
|
if(bytes_out > in_count) /* exit(2) if no savings */
|
365 |
|
|
exit_stat = 2;
|
366 |
|
|
printf("main: end...\n");
|
367 |
|
|
report (0xdeaddead);
|
368 |
|
|
return 0;
|
369 |
|
|
}
|
370 |
|
|
|
371 |
|
|
/*****************************************************************
|
372 |
|
|
* TAG( output )
|
373 |
|
|
*
|
374 |
|
|
* Output the given code.
|
375 |
|
|
* Inputs:
|
376 |
|
|
* code: A n_bits-bit integer. If == -1, then EOF. This assumes
|
377 |
|
|
* that n_bits =< (long)wordsize - 1.
|
378 |
|
|
* Outputs:
|
379 |
|
|
* Outputs code to the file.
|
380 |
|
|
* Assumptions:
|
381 |
|
|
* Chars are 8 bits long.
|
382 |
|
|
* Algorithm:
|
383 |
|
|
* Maintain a BITS character long buffer (so that 8 codes will
|
384 |
|
|
* fit in it exactly). Use the VAX insv instruction to insert each
|
385 |
|
|
* code in turn. When the buffer fills up empty it and start over.
|
386 |
|
|
*/
|
387 |
|
|
|
388 |
|
|
static char buf[BITS];
|
389 |
|
|
|
390 |
|
|
#ifndef vax
|
391 |
|
|
char_type lmask[9] = {0xff, 0xfe, 0xfc, 0xf8, 0xf0, 0xe0, 0xc0, 0x80, 0x00};
|
392 |
|
|
char_type rmask[9] = {0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff};
|
393 |
|
|
#endif /* vax */
|
394 |
|
|
|
395 |
|
|
void output( code )
|
396 |
|
|
code_int code;
|
397 |
|
|
{
|
398 |
|
|
|
399 |
|
|
/*
|
400 |
|
|
* On the VAX, it is important to have the register declarations
|
401 |
|
|
* in exactly the order given, or the asm will break.
|
402 |
|
|
*/
|
403 |
|
|
register int r_off = offset, bits= n_bits;
|
404 |
|
|
register char * bp = buf;
|
405 |
|
|
|
406 |
|
|
if ( code >= 0 ) {
|
407 |
|
|
#ifdef vax
|
408 |
|
|
/* VAX DEPENDENT!! Implementation on other machines is below.
|
409 |
|
|
*
|
410 |
|
|
* Translation: Insert BITS bits from the argument starting at
|
411 |
|
|
* offset bits from the beginning of buf.
|
412 |
|
|
*/
|
413 |
|
|
0; /* Work around for pcc -O bug with asm and if stmt */
|
414 |
|
|
asm( "insv 4(ap),r11,r10,(r9)" );
|
415 |
|
|
#else /* not a vax */
|
416 |
|
|
/*
|
417 |
|
|
* byte/bit numbering on the VAX is simulated by the following code
|
418 |
|
|
*/
|
419 |
|
|
/*
|
420 |
|
|
* Get to the first byte.
|
421 |
|
|
*/
|
422 |
|
|
bp += (r_off >> 3);
|
423 |
|
|
r_off &= 7;
|
424 |
|
|
/*
|
425 |
|
|
* Since code is always >= 8 bits, only need to mask the first
|
426 |
|
|
* hunk on the left.
|
427 |
|
|
*/
|
428 |
|
|
*bp = (*bp & rmask[r_off]) | ((code << r_off) & lmask[r_off]);
|
429 |
|
|
bp++;
|
430 |
|
|
bits -= (8 - r_off);
|
431 |
|
|
code >>= 8 - r_off;
|
432 |
|
|
/* Get any 8 bit parts in the middle (<=1 for up to 16 bits). */
|
433 |
|
|
if ( bits >= 8 ) {
|
434 |
|
|
*bp++ = code;
|
435 |
|
|
code >>= 8;
|
436 |
|
|
bits -= 8;
|
437 |
|
|
}
|
438 |
|
|
/* Last bits. */
|
439 |
|
|
if(bits)
|
440 |
|
|
*bp = code;
|
441 |
|
|
#endif /* vax */
|
442 |
|
|
offset += n_bits;
|
443 |
|
|
if ( offset == (n_bits << 3) ) {
|
444 |
|
|
bp = buf;
|
445 |
|
|
bits = n_bits;
|
446 |
|
|
bytes_out += bits;
|
447 |
|
|
/* do
|
448 |
|
|
putchar(*bp++); */
|
449 |
|
|
while(--bits);
|
450 |
|
|
offset = 0;
|
451 |
|
|
}
|
452 |
|
|
|
453 |
|
|
/*
|
454 |
|
|
* If the next entry is going to be too big for the code size,
|
455 |
|
|
* then increase it, if possible.
|
456 |
|
|
*/
|
457 |
|
|
if ( free_ent > maxcode || (clear_flg > 0))
|
458 |
|
|
{
|
459 |
|
|
/*
|
460 |
|
|
* Write the whole buffer, because the input side won't
|
461 |
|
|
* discover the size increase until after it has read it.
|
462 |
|
|
*/
|
463 |
|
|
if ( offset > 0 ) {
|
464 |
|
|
/* if( fwrite( buf, 1, n_bits, stdout ) != n_bits)
|
465 |
|
|
writeerr(); */
|
466 |
|
|
bytes_out += n_bits;
|
467 |
|
|
}
|
468 |
|
|
offset = 0;
|
469 |
|
|
|
470 |
|
|
if ( clear_flg ) {
|
471 |
|
|
maxcode = MAXCODE (n_bits = INIT_BITS);
|
472 |
|
|
clear_flg = 0;
|
473 |
|
|
}
|
474 |
|
|
else {
|
475 |
|
|
n_bits++;
|
476 |
|
|
if ( n_bits == maxbits )
|
477 |
|
|
maxcode = maxmaxcode;
|
478 |
|
|
else
|
479 |
|
|
maxcode = MAXCODE(n_bits);
|
480 |
|
|
}
|
481 |
|
|
}
|
482 |
|
|
} else {
|
483 |
|
|
/*
|
484 |
|
|
* At EOF, write the rest of the buffer.
|
485 |
|
|
*/
|
486 |
|
|
/* if ( offset > 0 )
|
487 |
|
|
fwrite( buf, 1, (offset + 7) / 8, stdout ); */
|
488 |
|
|
bytes_out += (offset + 7) / 8;
|
489 |
|
|
offset = 0;
|
490 |
|
|
/* fflush( stdout ); */
|
491 |
|
|
/* if( ferror( stdout ) )
|
492 |
|
|
writeerr(); */
|
493 |
|
|
}
|
494 |
|
|
}
|
495 |
|
|
|
496 |
|
|
/*
|
497 |
|
|
* Decompress stdin to stdout. This routine adapts to the codes in the
|
498 |
|
|
* file building the "string" table on-the-fly; requiring no table to
|
499 |
|
|
* be stored in the compressed file. The tables used herein are shared
|
500 |
|
|
* with those of the compress() routine. See the definitions above.
|
501 |
|
|
*/
|
502 |
|
|
|
503 |
|
|
void decompress() {
|
504 |
|
|
register char_type *stackp;
|
505 |
|
|
register int finchar;
|
506 |
|
|
register code_int code, oldcode, incode;
|
507 |
|
|
|
508 |
|
|
/*
|
509 |
|
|
* As above, initialize the first 256 entries in the table.
|
510 |
|
|
*/
|
511 |
|
|
maxcode = MAXCODE(n_bits = INIT_BITS);
|
512 |
|
|
for ( code = 255; code >= 0; code-- ) {
|
513 |
|
|
tab_prefixof(code) = 0;
|
514 |
|
|
tab_suffixof(code) = (char_type)code;
|
515 |
|
|
}
|
516 |
|
|
free_ent = ((block_compress) ? FIRST : 256 );
|
517 |
|
|
|
518 |
|
|
finchar = oldcode = getcode();
|
519 |
|
|
if(oldcode == -1) /* EOF already? */
|
520 |
|
|
return; /* Get out of here */
|
521 |
|
|
/* putchar( (char)finchar ); */ /* first code must be 8 bits = char */
|
522 |
|
|
/* if(ferror(stdout))
|
523 |
|
|
writeerr(); */
|
524 |
|
|
stackp = de_stack;
|
525 |
|
|
|
526 |
|
|
while ( (code = getcode()) > -1 ) {
|
527 |
|
|
|
528 |
|
|
if ( (code == CLEAR) && block_compress ) {
|
529 |
|
|
for ( code = 255; code >= 0; code-- )
|
530 |
|
|
tab_prefixof(code) = 0;
|
531 |
|
|
clear_flg = 1;
|
532 |
|
|
free_ent = FIRST - 1;
|
533 |
|
|
if ( (code = getcode ()) == -1 ) /* O, untimely death! */
|
534 |
|
|
break;
|
535 |
|
|
}
|
536 |
|
|
incode = code;
|
537 |
|
|
/*
|
538 |
|
|
* Special case for KwKwK string.
|
539 |
|
|
*/
|
540 |
|
|
if ( code >= free_ent ) {
|
541 |
|
|
*stackp++ = finchar;
|
542 |
|
|
code = oldcode;
|
543 |
|
|
}
|
544 |
|
|
|
545 |
|
|
/*
|
546 |
|
|
* Generate output characters in reverse order
|
547 |
|
|
*/
|
548 |
|
|
#ifdef SIGNED_COMPARE_SLOW
|
549 |
|
|
while ( ((unsigned long)code) >= ((unsigned long)256) ) {
|
550 |
|
|
#else
|
551 |
|
|
while ( code >= 256 ) {
|
552 |
|
|
#endif
|
553 |
|
|
*stackp++ = tab_suffixof(code);
|
554 |
|
|
code = tab_prefixof(code);
|
555 |
|
|
}
|
556 |
|
|
*stackp++ = finchar = tab_suffixof(code);
|
557 |
|
|
|
558 |
|
|
/*
|
559 |
|
|
* And put them out in forward order
|
560 |
|
|
*/
|
561 |
|
|
/* do
|
562 |
|
|
putchar ( *--stackp );
|
563 |
|
|
while ( stackp > de_stack );*/
|
564 |
|
|
|
565 |
|
|
/*
|
566 |
|
|
* Generate the new entry.
|
567 |
|
|
*/
|
568 |
|
|
if ( (code=free_ent) < maxmaxcode ) {
|
569 |
|
|
tab_prefixof(code) = (unsigned short)oldcode;
|
570 |
|
|
tab_suffixof(code) = finchar;
|
571 |
|
|
free_ent = code+1;
|
572 |
|
|
}
|
573 |
|
|
/*
|
574 |
|
|
* Remember previous code.
|
575 |
|
|
*/
|
576 |
|
|
oldcode = incode;
|
577 |
|
|
}
|
578 |
|
|
/* fflush( stdout ); */
|
579 |
|
|
/* if(ferror(stdout))
|
580 |
|
|
writeerr(); */
|
581 |
|
|
}
|
582 |
|
|
|
583 |
|
|
/*****************************************************************
|
584 |
|
|
* TAG( getcode )
|
585 |
|
|
*
|
586 |
|
|
* Read one code from the standard input. If EOF, return -1.
|
587 |
|
|
* Inputs:
|
588 |
|
|
* stdin
|
589 |
|
|
* Outputs:
|
590 |
|
|
* code or -1 is returned.
|
591 |
|
|
*/
|
592 |
|
|
|
593 |
|
|
code_int
|
594 |
|
|
getcode() {
|
595 |
|
|
/*
|
596 |
|
|
* On the VAX, it is important to have the register declarations
|
597 |
|
|
* in exactly the order given, or the asm will break.
|
598 |
|
|
*/
|
599 |
|
|
register code_int code;
|
600 |
|
|
static int offset = 0, size = 0;
|
601 |
|
|
static char_type buf[BITS];
|
602 |
|
|
register int r_off, bits;
|
603 |
|
|
register char_type *bp = buf;
|
604 |
|
|
|
605 |
|
|
if ( clear_flg > 0 || offset >= size || free_ent > maxcode ) {
|
606 |
|
|
/*
|
607 |
|
|
* If the next entry will be too big for the current code
|
608 |
|
|
* size, then we must increase the size. This implies reading
|
609 |
|
|
* a new buffer full, too.
|
610 |
|
|
*/
|
611 |
|
|
if ( free_ent > maxcode ) {
|
612 |
|
|
n_bits++;
|
613 |
|
|
if ( n_bits == maxbits )
|
614 |
|
|
maxcode = maxmaxcode; /* won't get any bigger now */
|
615 |
|
|
else
|
616 |
|
|
maxcode = MAXCODE(n_bits);
|
617 |
|
|
}
|
618 |
|
|
if ( clear_flg > 0) {
|
619 |
|
|
maxcode = MAXCODE (n_bits = INIT_BITS);
|
620 |
|
|
clear_flg = 0;
|
621 |
|
|
}
|
622 |
|
|
/* size = fread( buf, 1, n_bits, stdin ); */
|
623 |
|
|
if ( size <= 0 )
|
624 |
|
|
return -1; /* end of file */
|
625 |
|
|
offset = 0;
|
626 |
|
|
/* Round size down to integral number of codes */
|
627 |
|
|
size = (size << 3) - (n_bits - 1);
|
628 |
|
|
}
|
629 |
|
|
r_off = offset;
|
630 |
|
|
bits = n_bits;
|
631 |
|
|
#ifdef vax
|
632 |
|
|
asm( "extzv r10,r9,(r8),r11" );
|
633 |
|
|
#else /* not a vax */
|
634 |
|
|
/*
|
635 |
|
|
* Get to the first byte.
|
636 |
|
|
*/
|
637 |
|
|
bp += (r_off >> 3);
|
638 |
|
|
r_off &= 7;
|
639 |
|
|
/* Get first part (low order bits) */
|
640 |
|
|
#ifdef NO_UCHAR
|
641 |
|
|
code = ((*bp++ >> r_off) & rmask[8 - r_off]) & 0xff;
|
642 |
|
|
#else
|
643 |
|
|
code = (*bp++ >> r_off);
|
644 |
|
|
#endif /* NO_UCHAR */
|
645 |
|
|
bits -= (8 - r_off);
|
646 |
|
|
r_off = 8 - r_off; /* now, offset into code word */
|
647 |
|
|
/* Get any 8 bit parts in the middle (<=1 for up to 16 bits). */
|
648 |
|
|
if ( bits >= 8 ) {
|
649 |
|
|
#ifdef NO_UCHAR
|
650 |
|
|
code |= (*bp++ & 0xff) << r_off;
|
651 |
|
|
#else
|
652 |
|
|
code |= *bp++ << r_off;
|
653 |
|
|
#endif /* NO_UCHAR */
|
654 |
|
|
r_off += 8;
|
655 |
|
|
bits -= 8;
|
656 |
|
|
}
|
657 |
|
|
/* high order bits. */
|
658 |
|
|
code |= (*bp & rmask[bits]) << r_off;
|
659 |
|
|
#endif /* vax */
|
660 |
|
|
offset += n_bits;
|
661 |
|
|
|
662 |
|
|
return code;
|
663 |
|
|
}
|
664 |
|
|
|
665 |
|
|
/* For those who don't have it in libc.a */
|
666 |
|
|
char *
|
667 |
|
|
rindex(const char *s,
|
668 |
|
|
int c)
|
669 |
|
|
{
|
670 |
|
|
const char *p;
|
671 |
|
|
for (p = NULL; *s; s++)
|
672 |
|
|
if (*s == c)
|
673 |
|
|
p = s;
|
674 |
|
|
return((char *)p);
|
675 |
|
|
}
|
676 |
|
|
|
677 |
|
|
/*
|
678 |
|
|
writeerr()
|
679 |
|
|
{
|
680 |
|
|
perror ( ofname );
|
681 |
|
|
unlink ( ofname );
|
682 |
|
|
exit ( 1 );
|
683 |
|
|
}
|
684 |
|
|
*/
|
685 |
|
|
void cl_block () /* table clear for block compress */
|
686 |
|
|
{
|
687 |
|
|
register long int rat;
|
688 |
|
|
|
689 |
|
|
checkpoint = in_count + CHECK_GAP;
|
690 |
|
|
|
691 |
|
|
if(in_count > 0x007fffff) { /* shift will overflow */
|
692 |
|
|
rat = bytes_out >> 8;
|
693 |
|
|
if(rat == 0) { /* Don't divide by zero */
|
694 |
|
|
rat = 0x7fffffff;
|
695 |
|
|
} else {
|
696 |
|
|
rat = in_count / rat;
|
697 |
|
|
}
|
698 |
|
|
} else {
|
699 |
|
|
rat = (in_count << 8) / bytes_out; /* 8 fractional bits */
|
700 |
|
|
}
|
701 |
|
|
if ( rat > ratio ) {
|
702 |
|
|
ratio = rat;
|
703 |
|
|
} else {
|
704 |
|
|
ratio = 0;
|
705 |
|
|
cl_hash ( (count_int) hsize );
|
706 |
|
|
free_ent = FIRST;
|
707 |
|
|
clear_flg = 1;
|
708 |
|
|
output ( (code_int) CLEAR );
|
709 |
|
|
}
|
710 |
|
|
}
|
711 |
|
|
|
712 |
|
|
void cl_hash(hsize) /* reset code table */
|
713 |
|
|
register count_int hsize;
|
714 |
|
|
{
|
715 |
|
|
#ifndef XENIX_16 /* Normal machine */
|
716 |
|
|
register count_int *htab_p = htab+hsize;
|
717 |
|
|
#else
|
718 |
|
|
register j;
|
719 |
|
|
register long k = hsize;
|
720 |
|
|
register count_int *htab_p;
|
721 |
|
|
#endif
|
722 |
|
|
register long i;
|
723 |
|
|
register long m1 = -1;
|
724 |
|
|
|
725 |
|
|
#ifdef XENIX_16
|
726 |
|
|
for(j=0; j<=8 && k>=0; j++,k-=8192) {
|
727 |
|
|
i = 8192;
|
728 |
|
|
if(k < 8192) {
|
729 |
|
|
i = k;
|
730 |
|
|
}
|
731 |
|
|
htab_p = &(htab[j][i]);
|
732 |
|
|
i -= 16;
|
733 |
|
|
if(i > 0) {
|
734 |
|
|
#else
|
735 |
|
|
i = hsize - 16;
|
736 |
|
|
#endif
|
737 |
|
|
do { /* might use Sys V memset(3) here */
|
738 |
|
|
*(htab_p-16) = m1;
|
739 |
|
|
*(htab_p-15) = m1;
|
740 |
|
|
*(htab_p-14) = m1;
|
741 |
|
|
*(htab_p-13) = m1;
|
742 |
|
|
*(htab_p-12) = m1;
|
743 |
|
|
*(htab_p-11) = m1;
|
744 |
|
|
*(htab_p-10) = m1;
|
745 |
|
|
*(htab_p-9) = m1;
|
746 |
|
|
*(htab_p-8) = m1;
|
747 |
|
|
*(htab_p-7) = m1;
|
748 |
|
|
*(htab_p-6) = m1;
|
749 |
|
|
*(htab_p-5) = m1;
|
750 |
|
|
*(htab_p-4) = m1;
|
751 |
|
|
*(htab_p-3) = m1;
|
752 |
|
|
*(htab_p-2) = m1;
|
753 |
|
|
*(htab_p-1) = m1;
|
754 |
|
|
htab_p -= 16;
|
755 |
|
|
} while ((i -= 16) >= 0);
|
756 |
|
|
#ifdef XENIX_16
|
757 |
|
|
}
|
758 |
|
|
}
|
759 |
|
|
#endif
|
760 |
|
|
for ( i += 16; i > 0; i-- )
|
761 |
|
|
*--htab_p = m1;
|
762 |
|
|
}
|
763 |
|
|
|