1 /*
2 * compress - File compression ala IEEE Computer, June 1984.
3 *
4 * Algorithm from "A Technique for High Performance Data Compression",
5 * Terry A. Welch, IEEE Computer Vol 17, No 6 (June 1984), pp 8-19.
6 *
7 * Usage: compress [-dfvc] [-b bits] [file ...]
8 * Inputs:
9 * -b: limit the max number of bits/code.
10 * -c: write output on stdout, don't remove original.
11 * -d: decompress instead.
12 * -f: Forces output file to be generated, even if one already
13 * exists, and even if no space is saved by compressing.
14 * If -f is not used, the user will be prompted if stdin is
15 * a tty, otherwise, the output file will not be overwritten.
16 * -v: Write compression statistics
17 *
18 * file ...: Files to be compressed. If none specified, stdin is used.
19 * Outputs:
20 * file.Z: Compressed form of file with same mode, owner, and utimes
21 * or stdout (if stdin used as input)
22 *
23 * Assumptions:
24 * When filenames are given, replaces with the compressed version
25 * (.Z suffix) only if the file decreases in size.
26 * Algorithm:
27 * Modified Lempel-Ziv method (LZW). Basically finds common
28 * substrings and replaces them with a variable size code. This is
29 * deterministic, and can be done on the fly. Thus, the decompression
30 * procedure needs no input table, but tracks the way the table was built.
31
32 * Authors: Spencer W. Thomas (decvax!harpo!utah-cs!utah-gr!thomas)
33 * Jim McKie (decvax!mcvax!jim)
34 * Steve Davies (decvax!vax135!petsd!peora!srd)
35 * Ken Turkowski (decvax!decwrl!turtlevax!ken)
36 * James A. Woods (decvax!ihnp4!ames!jaw)
37 * Joe Orost (decvax!vax135!petsd!joe)
38 */
39
40 #ifndef USED
41 # define USED(x) if(x);else
42 #endif
43
44 #define _PLAN9_SOURCE
45 #define _BSD_EXTENSION
46 #define _POSIX_SOURCE
47
48 #include <u.h>
49 #include <stdio.h>
50 #include <ctype.h>
51 #include <stdlib.h>
52 #include <unistd.h>
53 #include <string.h>
54 #include <signal.h>
55 #include <utime.h>
56 #include <sys/types.h>
57 #include <sys/stat.h>
58
59 #define min(a,b) ((a>b) ? b : a)
60
61 #define BITS 16
62 #define HSIZE 69001 /* 95% occupancy */
63
64 /*
65 * a code_int must be able to hold 2**BITS values of type int, and also -1
66 */
67 typedef long code_int;
68 typedef long count_int;
69
70 static char rcs_ident[] = "$Header: compress.c,v 4.0 85/07/30 12:50:00 joe Release $";
71
72 uchar magic_header[] = { 0x1F, 0x9D }; /* 1F 9D */
73
74 /* Defines for third byte of header */
75 #define BIT_MASK 0x1f
76 #define BLOCK_MASK 0x80
77 /* Masks 0x40 and 0x20 are free. I think 0x20 should mean that there is
78 a fourth header byte (for expansion).
79 */
80 #define INIT_BITS 9 /* initial number of bits/code */
81
82 #define ARGVAL() (*++(*argv) || (--argc && *++argv))
83
84 int n_bits; /* number of bits/code */
85 int maxbits = BITS; /* user settable max # bits/code */
86 code_int maxcode; /* maximum code, given n_bits */
87 code_int maxmaxcode = 1 << BITS; /* should NEVER generate this code */
88
89 #define MAXCODE(n_bits) ((1 << (n_bits)) - 1)
90
91 count_int htab[HSIZE];
92 ushort codetab[HSIZE];
93
94 #define htabof(i) htab[i]
95 #define codetabof(i) codetab[i]
96
97 code_int hsize = HSIZE; /* for dynamic table sizing */
98 count_int fsize;
99
100 /*
101 * To save much memory, we overlay the table used by compress() with those
102 * used by decompress(). The tab_prefix table is the same size and type
103 * as the codetab. The tab_suffix table needs 2**BITS characters. We
104 * get this from the beginning of htab. The output stack uses the rest
105 * of htab, and contains characters. There is plenty of room for any
106 * possible stack (stack used to be 8000 characters).
107 */
108
109 #define tab_prefixof(i) codetabof(i)
110 #define tab_suffixof(i) ((uchar *)(htab))[i]
111 #define de_stack ((uchar *)&tab_suffixof(1<<BITS))
112
113 code_int free_ent = 0; /* first unused entry */
114 int exit_stat = 0;
115
116 void cl_block(void);
117 void cl_hash(count_int);
118 void compress(void);
119 void copystat(char *, char *);
120 void decompress(void);
121 int foreground(void);
122 code_int getcode(void);
123 void onintr(int);
124 void oops(int);
125 void output(code_int);
126 void prratio(FILE *, long, long);
127 void version(void);
128 void writeerr(void);
129
130 void
Usage(void)131 Usage(void)
132 {
133 #ifdef DEBUG
134 fprintf(stderr,"Usage: compress [-cdfDV] [-b maxbits] [file ...]\n");
135 #else
136 fprintf(stderr,"Usage: compress [-cdfvV] [-b maxbits] [file ...]\n");
137 #endif /* DEBUG */
138 }
139
140 int debug = 0;
141 int nomagic = 0; /* Use a 3-byte magic number header, unless old file */
142 int zcat_flg = 0; /* Write output on stdout, suppress messages */
143 int quiet = 1; /* don't tell me about compression */
144
145 /*
146 * block compression parameters -- after all codes are used up,
147 * and compression rate changes, start over.
148 */
149 int block_compress = BLOCK_MASK;
150 int clear_flg = 0;
151 long ratio = 0;
152 #define CHECK_GAP 10000 /* ratio check interval */
153 count_int checkpoint = CHECK_GAP;
154 /*
155 * the next two codes should not be changed lightly, as they must not
156 * lie within the contiguous general code space.
157 */
158 #define FIRST 257 /* first free entry */
159 #define CLEAR 256 /* table clear output code */
160
161 int force = 0;
162 char ofname [100];
163 #ifdef DEBUG
164 int verbose = 0;
165 #endif /* DEBUG */
166 void (*bgnd_flag)(int);
167
168 int do_decomp = 0;
169
170 int
main(argc,argv)171 main(argc, argv)
172 int argc;
173 char **argv;
174 {
175 int overwrite = 0; /* Do not overwrite unless given -f flag */
176 char tempname[512];
177 char **filelist, **fileptr;
178 char *cp;
179 struct stat statbuf;
180
181 if ( (bgnd_flag = signal ( SIGINT, SIG_IGN )) != SIG_IGN ) {
182 signal(SIGINT, onintr);
183 signal(SIGSEGV, oops);
184 }
185
186 filelist = fileptr = (char **)(malloc(argc * sizeof(*argv)));
187 *filelist = NULL;
188
189 if((cp = strrchr(argv[0], '/')) != 0)
190 cp++;
191 else
192 cp = argv[0];
193 if(strcmp(cp, "uncompress") == 0)
194 do_decomp = 1;
195 else if(strcmp(cp, "zcat") == 0) {
196 do_decomp = 1;
197 zcat_flg = 1;
198 }
199
200 /*
201 * Argument Processing
202 * All flags are optional.
203 * -C generate output compatible with compress 2.0.
204 * -D debug
205 * -V print Version; debug verbose
206 * -b maxbits maxbits. If -b is specified, then maxbits MUST be
207 * given also.
208 * -c cat all output to stdout
209 * -d do_decomp
210 * -f force overwrite of output file
211 * -n no header: useful to uncompress old files
212 * -v unquiet
213 * if a string is left, must be an input filename.
214 */
215 for (argc--, argv++; argc > 0; argc--, argv++) {
216 if (**argv == '-') { /* A flag argument */
217 while (*++(*argv)) { /* Process all flags in this arg */
218 switch (**argv) {
219 case 'C':
220 block_compress = 0;
221 break;
222 #ifdef DEBUG
223 case 'D':
224 debug = 1;
225 break;
226 case 'V':
227 verbose = 1;
228 version();
229 break;
230 #else
231 case 'V':
232 version();
233 break;
234 #endif
235 case 'b':
236 if (!ARGVAL()) {
237 fprintf(stderr, "Missing maxbits\n");
238 Usage();
239 exit(1);
240 }
241 maxbits = atoi(*argv);
242 goto nextarg;
243 case 'c':
244 zcat_flg = 1;
245 break;
246 case 'd':
247 do_decomp = 1;
248 break;
249 case 'f':
250 case 'F':
251 overwrite = 1;
252 force = 1;
253 break;
254 case 'n':
255 nomagic = 1;
256 break;
257 case 'q':
258 quiet = 1;
259 break;
260 case 'v':
261 quiet = 0;
262 break;
263 default:
264 fprintf(stderr, "Unknown flag: '%c'; ", **argv);
265 Usage();
266 exit(1);
267 }
268 }
269 } else { /* Input file name */
270 *fileptr++ = *argv; /* Build input file list */
271 *fileptr = NULL;
272 /* process nextarg; */
273 }
274 nextarg:
275 continue;
276 }
277
278 if(maxbits < INIT_BITS) maxbits = INIT_BITS;
279 if (maxbits > BITS) maxbits = BITS;
280 maxmaxcode = 1 << maxbits;
281
282 if (*filelist != NULL) {
283 for (fileptr = filelist; *fileptr; fileptr++) {
284 exit_stat = 0;
285 if (do_decomp != 0) { /* DECOMPRESSION */
286 /* Check for .Z suffix */
287 if (strcmp(*fileptr + strlen(*fileptr) - 2, ".Z") != 0) {
288 /* No .Z: tack one on */
289 strcpy(tempname, *fileptr);
290 strcat(tempname, ".Z");
291 *fileptr = tempname;
292 }
293 /* Open input file */
294 if ((freopen(*fileptr, "r", stdin)) == NULL) {
295 perror(*fileptr);
296 continue;
297 }
298 /* Check the magic number */
299 if (nomagic == 0) {
300 if ((getchar() != (magic_header[0] & 0xFF))
301 || (getchar() != (magic_header[1] & 0xFF))) {
302 fprintf(stderr, "%s: not in compressed format\n",
303 *fileptr);
304 continue;
305 }
306 maxbits = getchar(); /* set -b from file */
307 block_compress = maxbits & BLOCK_MASK;
308 maxbits &= BIT_MASK;
309 maxmaxcode = 1 << maxbits;
310 if(maxbits > BITS) {
311 fprintf(stderr,
312 "%s: compressed with %d bits, can only handle %d bits\n",
313 *fileptr, maxbits, BITS);
314 continue;
315 }
316 }
317 /* Generate output filename */
318 strcpy(ofname, *fileptr);
319 ofname[strlen(*fileptr) - 2] = '\0'; /* Strip off .Z */
320 } else { /* COMPRESSION */
321 if (strcmp(*fileptr + strlen(*fileptr) - 2, ".Z") == 0) {
322 fprintf(stderr,
323 "%s: already has .Z suffix -- no change\n",
324 *fileptr);
325 continue;
326 }
327 /* Open input file */
328 if ((freopen(*fileptr, "r", stdin)) == NULL) {
329 perror(*fileptr);
330 continue;
331 }
332 (void) stat(*fileptr, &statbuf);
333 fsize = (long) statbuf.st_size;
334 /*
335 * tune hash table size for small files -- ad hoc,
336 * but the sizes match earlier #defines, which
337 * serve as upper bounds on the number of output codes.
338 */
339 hsize = HSIZE;
340 if (fsize < (1 << 12))
341 hsize = min(5003, HSIZE);
342 else if (fsize < (1 << 13))
343 hsize = min(9001, HSIZE);
344 else if (fsize < (1 << 14))
345 hsize = min (18013, HSIZE);
346 else if (fsize < (1 << 15))
347 hsize = min (35023, HSIZE);
348 else if (fsize < 47000)
349 hsize = min (50021, HSIZE);
350
351 /* Generate output filename */
352 strcpy(ofname, *fileptr);
353 #ifndef BSD4_2
354 if ((cp=strrchr(ofname,'/')) != NULL)
355 cp++;
356 else
357 cp = ofname;
358 /*
359 *** changed 12 to 25; should be NAMELEN-3, but I don't want
360 * to fight the headers. ehg 5 Nov 92 **
361 */
362 if (strlen(cp) > 25) {
363 fprintf(stderr, "%s: filename too long to tack on .Z\n",
364 cp);
365 continue;
366 }
367 #endif
368 strcat(ofname, ".Z");
369 }
370 /* Check for overwrite of existing file */
371 if (overwrite == 0 && zcat_flg == 0 &&
372 stat(ofname, &statbuf) == 0) {
373 char response[2];
374
375 response[0] = 'n';
376 fprintf(stderr, "%s already exists;", ofname);
377 if (foreground()) {
378 fprintf(stderr,
379 " do you wish to overwrite %s (y or n)? ",
380 ofname);
381 fflush(stderr);
382 (void) read(2, response, 2);
383 while (response[1] != '\n')
384 if (read(2, response+1, 1) < 0) {
385 /* Ack! */
386 perror("stderr");
387 break;
388 }
389 }
390 if (response[0] != 'y') {
391 fprintf(stderr, "\tnot overwritten\n");
392 continue;
393 }
394 }
395 if(zcat_flg == 0) { /* Open output file */
396 if (freopen(ofname, "w", stdout) == NULL) {
397 perror(ofname);
398 continue;
399 }
400 if(!quiet)
401 fprintf(stderr, "%s: ", *fileptr);
402 }
403
404 /* Actually do the compression/decompression */
405 if (do_decomp == 0)
406 compress();
407 #ifndef DEBUG
408 else
409 decompress();
410 #else
411 else if (debug == 0)
412 decompress();
413 else
414 printcodes();
415 if (verbose)
416 dump_tab();
417 #endif /* DEBUG */
418 if(zcat_flg == 0) {
419 copystat(*fileptr, ofname); /* Copy stats */
420 if (exit_stat == 1 || !quiet)
421 putc('\n', stderr);
422 }
423 }
424 } else { /* Standard input */
425 if (do_decomp == 0) {
426 compress();
427 #ifdef DEBUG
428 if(verbose)
429 dump_tab();
430 #endif
431 if(!quiet)
432 putc('\n', stderr);
433 } else {
434 /* Check the magic number */
435 if (nomagic == 0) {
436 if ((getchar()!=(magic_header[0] & 0xFF))
437 || (getchar()!=(magic_header[1] & 0xFF))) {
438 fprintf(stderr, "stdin: not in compressed format\n");
439 exit(1);
440 }
441 maxbits = getchar(); /* set -b from file */
442 block_compress = maxbits & BLOCK_MASK;
443 maxbits &= BIT_MASK;
444 maxmaxcode = 1 << maxbits;
445 fsize = 100000; /* assume stdin large for USERMEM */
446 if(maxbits > BITS) {
447 fprintf(stderr,
448 "stdin: compressed with %d bits, can only handle %d bits\n",
449 maxbits, BITS);
450 exit(1);
451 }
452 }
453 #ifndef DEBUG
454 decompress();
455 #else
456 if (debug == 0)
457 decompress();
458 else
459 printcodes();
460 if (verbose)
461 dump_tab();
462 #endif /* DEBUG */
463 }
464 }
465 exit(exit_stat);
466 return 0;
467 }
468
469 static int offset;
470 long in_count = 1; /* length of input */
471 long bytes_out; /* length of compressed output */
472 long out_count = 0; /* # of codes output (for debugging) */
473
474 /*
475 * compress stdin to stdout
476 *
477 * Algorithm: use open addressing double hashing (no chaining) on the
478 * prefix code / next character combination. We do a variant of Knuth's
479 * algorithm D (vol. 3, sec. 6.4) along with G. Knott's relatively-prime
480 * secondary probe. Here, the modular division first probe is gives way
481 * to a faster exclusive-or manipulation. Also do block compression with
482 * an adaptive reset, whereby the code table is cleared when the compression
483 * ratio decreases, but after the table fills. The variable-length output
484 * codes are re-sized at this point, and a special CLEAR code is generated
485 * for the decompressor. Late addition: construct the table according to
486 * file size for noticeable speed improvement on small files. Please direct
487 * questions about this implementation to ames!jaw.
488 */
489 void
compress(void)490 compress(void)
491 {
492 code_int ent, hsize_reg;
493 code_int i;
494 int c, disp, hshift;
495 long fcode;
496
497 if (nomagic == 0) {
498 putchar(magic_header[0]);
499 putchar(magic_header[1]);
500 putchar((char)(maxbits | block_compress));
501 if(ferror(stdout))
502 writeerr();
503 }
504 offset = 0;
505 bytes_out = 3; /* includes 3-byte header mojo */
506 out_count = 0;
507 clear_flg = 0;
508 ratio = 0;
509 in_count = 1;
510 checkpoint = CHECK_GAP;
511 maxcode = MAXCODE(n_bits = INIT_BITS);
512 free_ent = (block_compress? FIRST: 256);
513
514 ent = getchar ();
515
516 hshift = 0;
517 for (fcode = (long)hsize; fcode < 65536L; fcode *= 2)
518 hshift++;
519 hshift = 8 - hshift; /* set hash code range bound */
520
521 hsize_reg = hsize;
522 cl_hash( (count_int) hsize_reg); /* clear hash table */
523
524 while ((c = getchar()) != EOF) {
525 in_count++;
526 fcode = (long) (((long) c << maxbits) + ent);
527 i = ((c << hshift) ^ ent); /* xor hashing */
528
529 if (htabof (i) == fcode) {
530 ent = codetabof(i);
531 continue;
532 } else if ((long)htabof(i) < 0 ) /* empty slot */
533 goto nomatch;
534 disp = hsize_reg - i; /* secondary hash (after G. Knott) */
535 if (i == 0)
536 disp = 1;
537 probe:
538 if ((i -= disp) < 0)
539 i += hsize_reg;
540
541 if (htabof (i) == fcode) {
542 ent = codetabof(i);
543 continue;
544 }
545 if ((long)htabof(i) > 0)
546 goto probe;
547 nomatch:
548 output((code_int)ent);
549 out_count++;
550 ent = c;
551 if (free_ent < maxmaxcode) {
552 codetabof(i) = free_ent++; /* code -> hashtable */
553 htabof(i) = fcode;
554 } else if ((count_int)in_count >= checkpoint && block_compress)
555 cl_block ();
556 }
557 /*
558 * Put out the final code.
559 */
560 output( (code_int)ent );
561 out_count++;
562 output( (code_int)-1 );
563
564 /*
565 * Print out stats on stderr
566 */
567 if(zcat_flg == 0 && !quiet) {
568 #ifdef DEBUG
569 fprintf( stderr,
570 "%ld chars in, %ld codes (%ld bytes) out, compression factor: ",
571 in_count, out_count, bytes_out );
572 prratio( stderr, in_count, bytes_out );
573 fprintf( stderr, "\n");
574 fprintf( stderr, "\tCompression as in compact: " );
575 prratio( stderr, in_count-bytes_out, in_count );
576 fprintf( stderr, "\n");
577 fprintf( stderr, "\tLargest code (of last block) was %d (%d bits)\n",
578 free_ent - 1, n_bits );
579 #else /* !DEBUG */
580 fprintf( stderr, "Compression: " );
581 prratio( stderr, in_count-bytes_out, in_count );
582 #endif /* DEBUG */
583 }
584 if(bytes_out > in_count) /* exit(2) if no savings */
585 exit_stat = 2;
586 }
587
588 /*
589 * TAG( output )
590 *
591 * Output the given code.
592 * Inputs:
593 * code: A n_bits-bit integer. If == -1, then EOF. This assumes
594 * that n_bits =< (long)wordsize - 1.
595 * Outputs:
596 * Outputs code to the file.
597 * Assumptions:
598 * Chars are 8 bits long.
599 * Algorithm:
600 * Maintain a BITS character long buffer (so that 8 codes will
601 * fit in it exactly). When the buffer fills up empty it and start over.
602 */
603
604 static char buf[BITS];
605
606 uchar lmask[9] = {0xff, 0xfe, 0xfc, 0xf8, 0xf0, 0xe0, 0xc0, 0x80, 0x00};
607 uchar rmask[9] = {0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff};
608
609 void
output(code)610 output( code )
611 code_int code;
612 {
613 #ifdef DEBUG
614 static int col = 0;
615 #endif
616 int r_off = offset, bits= n_bits;
617 char *bp = buf;
618
619 #ifdef DEBUG
620 if (verbose)
621 fprintf(stderr, "%5d%c", code,
622 (col+=6) >= 74? (col = 0, '\n'): ' ');
623 #endif
624 if (code >= 0) {
625 /*
626 * byte/bit numbering on the VAX is simulated by the
627 * following code
628 */
629 /*
630 * Get to the first byte.
631 */
632 bp += (r_off >> 3);
633 r_off &= 7;
634 /*
635 * Since code is always >= 8 bits, only need to mask the first
636 * hunk on the left.
637 */
638 *bp = (*bp & rmask[r_off]) | (code << r_off) & lmask[r_off];
639 bp++;
640 bits -= 8 - r_off;
641 code >>= 8 - r_off;
642 /* Get any 8 bit parts in the middle (<=1 for up to 16 bits). */
643 if ( bits >= 8 ) {
644 *bp++ = code;
645 code >>= 8;
646 bits -= 8;
647 }
648 /* Last bits. */
649 if(bits)
650 *bp = code;
651
652 offset += n_bits;
653 if ( offset == (n_bits << 3) ) {
654 bp = buf;
655 bits = n_bits;
656 bytes_out += bits;
657 do {
658 putchar(*bp++);
659 } while(--bits);
660 offset = 0;
661 }
662
663 /*
664 * If the next entry is going to be too big for the code size,
665 * then increase it, if possible.
666 */
667 if ( free_ent > maxcode || (clear_flg > 0)) {
668 /*
669 * Write the whole buffer, because the input side won't
670 * discover the size increase until after it has read it.
671 */
672 if ( offset > 0 ) {
673 if( fwrite( buf, 1, n_bits, stdout ) != n_bits)
674 writeerr();
675 bytes_out += n_bits;
676 }
677 offset = 0;
678
679 if ( clear_flg ) {
680 maxcode = MAXCODE (n_bits = INIT_BITS);
681 clear_flg = 0;
682 } else {
683 n_bits++;
684 if ( n_bits == maxbits )
685 maxcode = maxmaxcode;
686 else
687 maxcode = MAXCODE(n_bits);
688 }
689 #ifdef DEBUG
690 if ( debug ) {
691 fprintf(stderr,
692 "\nChange to %d bits\n", n_bits);
693 col = 0;
694 }
695 #endif
696 }
697 } else {
698 /*
699 * At EOF, write the rest of the buffer.
700 */
701 if ( offset > 0 )
702 fwrite( buf, 1, (offset + 7) / 8, stdout );
703 bytes_out += (offset + 7) / 8;
704 offset = 0;
705 fflush( stdout );
706 #ifdef DEBUG
707 if ( verbose )
708 fprintf( stderr, "\n" );
709 #endif
710 if( ferror( stdout ) )
711 writeerr();
712 }
713 }
714
715 /*
716 * Decompress stdin to stdout. This routine adapts to the codes in the
717 * file building the "string" table on-the-fly; requiring no table to
718 * be stored in the compressed file. The tables used herein are shared
719 * with those of the compress() routine. See the definitions above.
720 */
721 void
decompress(void)722 decompress(void)
723 {
724 int finchar;
725 code_int code, oldcode, incode;
726 uchar *stackp;
727
728 /*
729 * As above, initialize the first 256 entries in the table.
730 */
731 maxcode = MAXCODE(n_bits = INIT_BITS);
732 for (code = 255; code >= 0; code--) {
733 tab_prefixof(code) = 0;
734 tab_suffixof(code) = (uchar)code;
735 }
736 free_ent = (block_compress? FIRST: 256);
737
738 finchar = oldcode = getcode();
739 if(oldcode == -1) /* EOF already? */
740 return; /* Get out of here */
741 putchar((char)finchar); /* first code must be 8 bits = char */
742 if(ferror(stdout)) /* Crash if can't write */
743 writeerr();
744 stackp = de_stack;
745
746 while ((code = getcode()) > -1) {
747 if ((code == CLEAR) && block_compress) {
748 for (code = 255; code >= 0; code--)
749 tab_prefixof(code) = 0;
750 clear_flg = 1;
751 free_ent = FIRST - 1;
752 if ((code = getcode()) == -1) /* O, untimely death! */
753 break;
754 }
755 incode = code;
756 /*
757 * Special case for KwKwK string.
758 */
759 if (code >= free_ent) {
760 *stackp++ = finchar;
761 code = oldcode;
762 }
763
764 /*
765 * Generate output characters in reverse order
766 */
767 while (code >= 256) {
768 *stackp++ = tab_suffixof(code);
769 code = tab_prefixof(code);
770 }
771 *stackp++ = finchar = tab_suffixof(code);
772
773 /*
774 * And put them out in forward order
775 */
776 do {
777 putchar(*--stackp);
778 } while (stackp > de_stack);
779
780 /*
781 * Generate the new entry.
782 */
783 if ( (code=free_ent) < maxmaxcode ) {
784 tab_prefixof(code) = (ushort)oldcode;
785 tab_suffixof(code) = finchar;
786 free_ent = code+1;
787 }
788 /*
789 * Remember previous code.
790 */
791 oldcode = incode;
792 }
793 fflush(stdout);
794 if(ferror(stdout))
795 writeerr();
796 }
797
798 /*
799 * TAG( getcode )
800 *
801 * Read one code from the standard input. If EOF, return -1.
802 * Inputs:
803 * stdin
804 * Outputs:
805 * code or -1 is returned.
806 */
807 code_int
getcode()808 getcode()
809 {
810 int r_off, bits;
811 code_int code;
812 static int offset = 0, size = 0;
813 static uchar buf[BITS];
814 uchar *bp = buf;
815
816 if ( clear_flg > 0 || offset >= size || free_ent > maxcode ) {
817 /*
818 * If the next entry will be too big for the current code
819 * size, then we must increase the size. This implies reading
820 * a new buffer full, too.
821 */
822 if ( free_ent > maxcode ) {
823 n_bits++;
824 if ( n_bits == maxbits )
825 maxcode = maxmaxcode; /* won't get any bigger now */
826 else
827 maxcode = MAXCODE(n_bits);
828 }
829 if ( clear_flg > 0) {
830 maxcode = MAXCODE(n_bits = INIT_BITS);
831 clear_flg = 0;
832 }
833 size = fread(buf, 1, n_bits, stdin);
834 if (size <= 0)
835 return -1; /* end of file */
836 offset = 0;
837 /* Round size down to integral number of codes */
838 size = (size << 3) - (n_bits - 1);
839 }
840 r_off = offset;
841 bits = n_bits;
842 /*
843 * Get to the first byte.
844 */
845 bp += (r_off >> 3);
846 r_off &= 7;
847 /* Get first part (low order bits) */
848 code = (*bp++ >> r_off);
849 bits -= (8 - r_off);
850 r_off = 8 - r_off; /* now, offset into code word */
851 /* Get any 8 bit parts in the middle (<=1 for up to 16 bits). */
852 if (bits >= 8) {
853 code |= *bp++ << r_off;
854 r_off += 8;
855 bits -= 8;
856 }
857 /* high order bits. */
858 code |= (*bp & rmask[bits]) << r_off;
859 offset += n_bits;
860 return code;
861 }
862
863 #ifdef DEBUG
printcodes()864 printcodes()
865 {
866 /*
867 * Just print out codes from input file. For debugging.
868 */
869 code_int code;
870 int col = 0, bits;
871
872 bits = n_bits = INIT_BITS;
873 maxcode = MAXCODE(n_bits);
874 free_ent = ((block_compress) ? FIRST : 256 );
875 while ( ( code = getcode() ) >= 0 ) {
876 if ( (code == CLEAR) && block_compress ) {
877 free_ent = FIRST - 1;
878 clear_flg = 1;
879 }
880 else if ( free_ent < maxmaxcode )
881 free_ent++;
882 if ( bits != n_bits ) {
883 fprintf(stderr, "\nChange to %d bits\n", n_bits );
884 bits = n_bits;
885 col = 0;
886 }
887 fprintf(stderr, "%5d%c", code, (col+=6) >= 74 ? (col = 0, '\n') : ' ' );
888 }
889 putc( '\n', stderr );
890 exit( 0 );
891 }
892
893 code_int sorttab[1<<BITS]; /* sorted pointers into htab */
894
895 #define STACK_SIZE 15000
896
dump_tab()897 dump_tab() /* dump string table */
898 {
899 int i, first, c, ent;
900 int stack_top = STACK_SIZE;
901
902 if(do_decomp == 0) { /* compressing */
903 int flag = 1;
904
905 for(i=0; i<hsize; i++) { /* build sort pointers */
906 if((long)htabof(i) >= 0) {
907 sorttab[codetabof(i)] = i;
908 }
909 }
910 first = block_compress ? FIRST : 256;
911 for(i = first; i < free_ent; i++) {
912 fprintf(stderr, "%5d: \"", i);
913 de_stack[--stack_top] = '\n';
914 de_stack[--stack_top] = '"';
915 stack_top = in_stack((htabof(sorttab[i])>>maxbits)&0xff,
916 stack_top);
917 for(ent=htabof(sorttab[i]) & ((1<<maxbits)-1);
918 ent > 256;
919 ent=htabof(sorttab[ent]) & ((1<<maxbits)-1)) {
920 stack_top = in_stack(htabof(sorttab[ent]) >> maxbits,
921 stack_top);
922 }
923 stack_top = in_stack(ent, stack_top);
924 fwrite( &de_stack[stack_top], 1, STACK_SIZE-stack_top, stderr);
925 stack_top = STACK_SIZE;
926 }
927 } else if(!debug) { /* decompressing */
928
929 for ( i = 0; i < free_ent; i++ ) {
930 ent = i;
931 c = tab_suffixof(ent);
932 if ( isascii(c) && isprint(c) )
933 fprintf( stderr, "%5d: %5d/'%c' \"",
934 ent, tab_prefixof(ent), c );
935 else
936 fprintf( stderr, "%5d: %5d/\\%03o \"",
937 ent, tab_prefixof(ent), c );
938 de_stack[--stack_top] = '\n';
939 de_stack[--stack_top] = '"';
940 for ( ; ent != NULL;
941 ent = (ent >= FIRST ? tab_prefixof(ent) : NULL) ) {
942 stack_top = in_stack(tab_suffixof(ent), stack_top);
943 }
944 fwrite( &de_stack[stack_top], 1, STACK_SIZE - stack_top, stderr );
945 stack_top = STACK_SIZE;
946 }
947 }
948 }
949
950 int
in_stack(int c,int stack_top)951 in_stack(int c, int stack_top)
952 {
953 if ( (isascii(c) && isprint(c) && c != '\\') || c == ' ' ) {
954 de_stack[--stack_top] = c;
955 } else {
956 switch( c ) {
957 case '\n': de_stack[--stack_top] = 'n'; break;
958 case '\t': de_stack[--stack_top] = 't'; break;
959 case '\b': de_stack[--stack_top] = 'b'; break;
960 case '\f': de_stack[--stack_top] = 'f'; break;
961 case '\r': de_stack[--stack_top] = 'r'; break;
962 case '\\': de_stack[--stack_top] = '\\'; break;
963 default:
964 de_stack[--stack_top] = '0' + c % 8;
965 de_stack[--stack_top] = '0' + (c / 8) % 8;
966 de_stack[--stack_top] = '0' + c / 64;
967 break;
968 }
969 de_stack[--stack_top] = '\\';
970 }
971 return stack_top;
972 }
973 #endif /* DEBUG */
974
975 void
writeerr(void)976 writeerr(void)
977 {
978 perror(ofname);
979 unlink(ofname);
980 exit(1);
981 }
982
983 void
copystat(ifname,ofname)984 copystat(ifname, ofname)
985 char *ifname, *ofname;
986 {
987 int mode;
988 time_t timep[2]; /* should be struct utimbuf */
989 struct stat statbuf;
990
991 fclose(stdout);
992 if (stat(ifname, &statbuf)) { /* Get stat on input file */
993 perror(ifname);
994 return;
995 }
996 if (!S_ISREG(statbuf.st_mode)) {
997 if (quiet)
998 fprintf(stderr, "%s: ", ifname);
999 fprintf(stderr, " -- not a regular file: unchanged");
1000 exit_stat = 1;
1001 } else if (exit_stat == 2 && !force) {
1002 /* No compression: remove file.Z */
1003 if (!quiet)
1004 fprintf(stderr, " -- file unchanged");
1005 } else { /* Successful Compression */
1006 exit_stat = 0;
1007 mode = statbuf.st_mode & 0777;
1008 if (chmod(ofname, mode)) /* Copy modes */
1009 perror(ofname);
1010 /* Copy ownership */
1011 chown(ofname, statbuf.st_uid, statbuf.st_gid);
1012 timep[0] = statbuf.st_atime;
1013 timep[1] = statbuf.st_mtime;
1014 /* Update last accessed and modified times */
1015 utime(ofname, (struct utimbuf *)timep);
1016 // if (unlink(ifname)) /* Remove input file */
1017 // perror(ifname);
1018 return; /* success */
1019 }
1020
1021 /* Unsuccessful return -- one of the tests failed */
1022 if (unlink(ofname))
1023 perror(ofname);
1024 }
1025
1026 /*
1027 * This routine returns 1 if we are running in the foreground and stderr
1028 * is a tty.
1029 */
1030 int
foreground(void)1031 foreground(void)
1032 {
1033 if(bgnd_flag) /* background? */
1034 return 0;
1035 else /* foreground */
1036 return isatty(2); /* and stderr is a tty */
1037 }
1038
1039 void
onintr(int x)1040 onintr(int x)
1041 {
1042 USED(x);
1043 unlink(ofname);
1044 exit(1);
1045 }
1046
1047 void
oops(int x)1048 oops(int x) /* wild pointer -- assume bad input */
1049 {
1050 USED(x);
1051 if (do_decomp == 1)
1052 fprintf(stderr, "uncompress: corrupt input\n");
1053 unlink(ofname);
1054 exit(1);
1055 }
1056
1057 void
cl_block(void)1058 cl_block(void) /* table clear for block compress */
1059 {
1060 long rat;
1061
1062 checkpoint = in_count + CHECK_GAP;
1063 #ifdef DEBUG
1064 if ( debug ) {
1065 fprintf ( stderr, "count: %ld, ratio: ", in_count );
1066 prratio ( stderr, in_count, bytes_out );
1067 fprintf ( stderr, "\n");
1068 }
1069 #endif /* DEBUG */
1070
1071 if (in_count > 0x007fffff) { /* shift will overflow */
1072 rat = bytes_out >> 8;
1073 if (rat == 0) /* Don't divide by zero */
1074 rat = 0x7fffffff;
1075 else
1076 rat = in_count / rat;
1077 } else
1078 rat = (in_count << 8) / bytes_out; /* 8 fractional bits */
1079 if (rat > ratio)
1080 ratio = rat;
1081 else {
1082 ratio = 0;
1083 #ifdef DEBUG
1084 if (verbose)
1085 dump_tab(); /* dump string table */
1086 #endif
1087 cl_hash((count_int)hsize);
1088 free_ent = FIRST;
1089 clear_flg = 1;
1090 output((code_int)CLEAR);
1091 #ifdef DEBUG
1092 if (debug)
1093 fprintf(stderr, "clear\n");
1094 #endif /* DEBUG */
1095 }
1096 }
1097
1098 void
cl_hash(count_int hsize)1099 cl_hash(count_int hsize) /* reset code table */
1100 {
1101 count_int *htab_p = htab+hsize;
1102 long i;
1103 long m1 = -1;
1104
1105 i = hsize - 16;
1106 do { /* might use Sys V memset(3) here */
1107 *(htab_p-16) = m1;
1108 *(htab_p-15) = m1;
1109 *(htab_p-14) = m1;
1110 *(htab_p-13) = m1;
1111 *(htab_p-12) = m1;
1112 *(htab_p-11) = m1;
1113 *(htab_p-10) = m1;
1114 *(htab_p-9) = m1;
1115 *(htab_p-8) = m1;
1116 *(htab_p-7) = m1;
1117 *(htab_p-6) = m1;
1118 *(htab_p-5) = m1;
1119 *(htab_p-4) = m1;
1120 *(htab_p-3) = m1;
1121 *(htab_p-2) = m1;
1122 *(htab_p-1) = m1;
1123 htab_p -= 16;
1124 } while ((i -= 16) >= 0);
1125 for ( i += 16; i > 0; i-- )
1126 *--htab_p = m1;
1127 }
1128
1129 void
prratio(stream,num,den)1130 prratio(stream, num, den)
1131 FILE *stream;
1132 long num, den;
1133 {
1134 int q; /* Doesn't need to be long */
1135
1136 if(num > 214748L) /* 2147483647/10000 */
1137 q = num / (den / 10000L);
1138 else
1139 q = 10000L * num / den; /* Long calculations, though */
1140 if (q < 0) {
1141 putc('-', stream);
1142 q = -q;
1143 }
1144 fprintf(stream, "%d.%02d%%", q / 100, q % 100);
1145 }
1146
1147 void
version(void)1148 version(void)
1149 {
1150 fprintf(stderr, "%s\n", rcs_ident);
1151 fprintf(stderr, "Options: ");
1152 #ifdef DEBUG
1153 fprintf(stderr, "DEBUG, ");
1154 #endif
1155 #ifdef BSD4_2
1156 fprintf(stderr, "BSD4_2, ");
1157 #endif
1158 fprintf(stderr, "BITS = %d\n", BITS);
1159 }
1160
1161 /*
1162 * The revision-history novel:
1163 *
1164 * $Header: compress.c,v 4.0 85/07/30 12:50:00 joe Release $
1165 * $Log: compress.c,v $
1166 * Revision 4.0 85/07/30 12:50:00 joe
1167 * Removed ferror() calls in output routine on every output except first.
1168 * Prepared for release to the world.
1169 *
1170 * Revision 3.6 85/07/04 01:22:21 joe
1171 * Remove much wasted storage by overlaying hash table with the tables
1172 * used by decompress: tab_suffix[1<<BITS], stack[8000]. Updated USERMEM
1173 * computations. Fixed dump_tab() DEBUG routine.
1174 *
1175 * Revision 3.5 85/06/30 20:47:21 jaw
1176 * Change hash function to use exclusive-or. Rip out hash cache. These
1177 * speedups render the megamemory version defunct, for now. Make decoder
1178 * stack global. Parts of the RCS trunks 2.7, 2.6, and 2.1 no longer apply.
1179 *
1180 * Revision 3.4 85/06/27 12:00:00 ken
1181 * Get rid of all floating-point calculations by doing all compression ratio
1182 * calculations in fixed point.
1183 *
1184 * Revision 3.3 85/06/24 21:53:24 joe
1185 * Incorporate portability suggestion for M_XENIX. Got rid of text on #else
1186 * and #endif lines. Cleaned up #ifdefs for vax and interdata.
1187 *
1188 * Revision 3.2 85/06/06 21:53:24 jaw
1189 * Incorporate portability suggestions for Z8000, IBM PC/XT from mailing list.
1190 * Default to "quiet" output (no compression statistics).
1191 *
1192 * Revision 3.1 85/05/12 18:56:13 jaw
1193 * Integrate decompress() stack speedups (from early pointer mods by McKie).
1194 * Repair multi-file USERMEM gaffe. Unify 'force' flags to mimic semantics
1195 * of SVR2 'pack'. Streamline block-compress table clear logic. Increase
1196 * output byte count by magic number size.
1197 *
1198 * Revision 3.0 84/11/27 11:50:00 petsd!joe
1199 * Set HSIZE depending on BITS. Set BITS depending on USERMEM. Unrolled
1200 * loops in clear routines. Added "-C" flag for 2.0 compatibility. Used
1201 * unsigned compares on Perkin-Elmer. Fixed foreground check.
1202 *
1203 * Revision 2.7 84/11/16 19:35:39 ames!jaw
1204 * Cache common hash codes based on input statistics; this improves
1205 * performance for low-density raster images. Pass on #ifdef bundle
1206 * from Turkowski.
1207 *
1208 * Revision 2.6 84/11/05 19:18:21 ames!jaw
1209 * Vary size of hash tables to reduce time for small files.
1210 * Tune PDP-11 hash function.
1211 *
1212 * Revision 2.5 84/10/30 20:15:14 ames!jaw
1213 * Junk chaining; replace with the simpler (and, on the VAX, faster)
1214 * double hashing, discussed within. Make block compression standard.
1215 *
1216 * Revision 2.4 84/10/16 11:11:11 ames!jaw
1217 * Introduce adaptive reset for block compression, to boost the rate
1218 * another several percent. (See mailing list notes.)
1219 *
1220 * Revision 2.3 84/09/22 22:00:00 petsd!joe
1221 * Implemented "-B" block compress. Implemented REVERSE sorting of tab_next.
1222 * Bug fix for last bits. Changed fwrite to putchar loop everywhere.
1223 *
1224 * Revision 2.2 84/09/18 14:12:21 ames!jaw
1225 * Fold in news changes, small machine typedef from thomas,
1226 * #ifdef interdata from joe.
1227 *
1228 * Revision 2.1 84/09/10 12:34:56 ames!jaw
1229 * Configured fast table lookup for 32-bit machines.
1230 * This cuts user time in half for b <= FBITS, and is useful for news batching
1231 * from VAX to PDP sites. Also sped up decompress() [fwrite->putc] and
1232 * added signal catcher [plus beef in writeerr()] to delete effluvia.
1233 *
1234 * Revision 2.0 84/08/28 22:00:00 petsd!joe
1235 * Add check for foreground before prompting user. Insert maxbits into
1236 * compressed file. Force file being uncompressed to end with ".Z".
1237 * Added "-c" flag and "zcat". Prepared for release.
1238 *
1239 * Revision 1.10 84/08/24 18:28:00 turtlevax!ken
1240 * Will only compress regular files (no directories), added a magic number
1241 * header (plus an undocumented -n flag to handle old files without headers),
1242 * added -f flag to force overwriting of possibly existing destination file,
1243 * otherwise the user is prompted for a response. Will tack on a .Z to a
1244 * filename if it doesn't have one when decompressing. Will only replace
1245 * file if it was compressed.
1246 *
1247 * Revision 1.9 84/08/16 17:28:00 turtlevax!ken
1248 * Removed scanargs(), getopt(), added .Z extension and unlimited number of
1249 * filenames to compress. Flags may be clustered (-Ddvb12) or separated
1250 * (-D -d -v -b 12), or combination thereof. Modes and other status is
1251 * copied with copystat(). -O bug for 4.2 seems to have disappeared with
1252 * 1.8.
1253 *
1254 * Revision 1.8 84/08/09 23:15:00 joe
1255 * Made it compatible with vax version, installed jim's fixes/enhancements
1256 *
1257 * Revision 1.6 84/08/01 22:08:00 joe
1258 * Sped up algorithm significantly by sorting the compress chain.
1259 *
1260 * Revision 1.5 84/07/13 13:11:00 srd
1261 * Added C version of vax asm routines. Changed structure to arrays to
1262 * save much memory. Do unsigned compares where possible (faster on
1263 * Perkin-Elmer)
1264 *
1265 * Revision 1.4 84/07/05 03:11:11 thomas
1266 * Clean up the code a little and lint it. (Lint complains about all
1267 * the regs used in the asm, but I'm not going to "fix" this.)
1268 *
1269 * Revision 1.3 84/07/05 02:06:54 thomas
1270 * Minor fixes.
1271 *
1272 * Revision 1.2 84/07/05 00:27:27 thomas
1273 * Add variable bit length output.
1274 */
1275