![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to compress.c CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [MW Coherent from dump] / coherent / b / bin / compress|
Return to compress.c CVS log | Up to [MW Coherent from dump] / coherent / b / bin / compress |
1.1 root 1: #ifdef SCCSID
2: static char *SccsId = "@(#)compress.c 1.14 9/24/87";
3: #endif /* SCCSID */
4: static char rcs_ident[] = "Based on compress.c,v 4.0 85/07/30 12:50:00 joe Release";
5:
6: /*
7: * Compress - data compression program
8: */
9: #define min(a,b) ((a>b) ? b : a)
10:
11: /*
12: * machine variants which require cc -Dmachine: pdp11, z8000, pcxt
13: */
14:
15: /*
16: * Set USERMEM to the maximum amount of physical user memory available
17: * in bytes. USERMEM is used to determine the maximum BITS that can be used
18: * for compression.
19: *
20: * SACREDMEM is the amount of physical memory saved for others; compress
21: * will hog the rest.
22: */
23: #ifndef SACREDMEM
24: #define SACREDMEM 0
25: #endif
26:
27: #ifndef USERMEM
28: # define USERMEM 450000 /* default user memory */
29: #endif
30:
31: #ifdef interdata /* (Perkin-Elmer) */
32: #define SIGNED_COMPARE_SLOW /* signed compare is slower than unsigned */
33: #endif
34:
35: #ifdef pdp11
36: # define BITS 12 /* max bits/code for 16-bit machine */
37: # define NO_UCHAR /* also if "unsigned char" functions as signed char */
38: # undef USERMEM
39: #endif /* pdp11 */ /* don't forget to compile with -i */
40:
41: #ifdef z8000
42: # define BITS 12
43: # undef vax /* weird preprocessor */
44: # undef USERMEM
45: #endif /* z8000 */
46:
47: #ifdef pcxt
48: # define BITS 12
49: # undef USERMEM
50: #endif /* pcxt */
51:
52: #ifdef USERMEM
53: # if USERMEM >= (433484+SACREDMEM)
54: # define PBITS 16
55: # else
56: # if USERMEM >= (229600+SACREDMEM)
57: # define PBITS 15
58: # else
59: # if USERMEM >= (127536+SACREDMEM)
60: # define PBITS 14
61: # else
62: # if USERMEM >= (73464+SACREDMEM)
63: # define PBITS 13
64: # else
65: # define PBITS 12
66: # endif
67: # endif
68: # endif
69: # endif
70: # undef USERMEM
71: #endif /* USERMEM */
72:
73: #ifdef PBITS /* Preferred BITS for this memory size */
74: # ifndef BITS
75: # define BITS PBITS
76: # endif /* BITS */
77: #endif /* PBITS */
78:
79: #if BITS == 16
80: # define HSIZE 69001 /* 95% occupancy */
81: #endif
82: #if BITS == 15
83: # define HSIZE 35023 /* 94% occupancy */
84: #endif
85: #if BITS == 14
86: # define HSIZE 18013 /* 91% occupancy */
87: #endif
88: #if BITS == 13
89: # define HSIZE 9001 /* 91% occupancy */
90: #endif
91: #if BITS <= 12
92: # define HSIZE 5003 /* 80% occupancy */
93: #endif
94:
95: #ifdef M_XENIX /* Stupid compiler can't handle arrays with */
96: # if BITS == 16 /* more than 65535 bytes - so we fake it */
97: # define XENIX_16
98: # else
99: # if BITS > 13 /* Code only handles BITS = 12, 13, or 16 */
100: # define BITS 13
101: # endif
102: # endif
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: #include <stdio.h>
149: #include <ctype.h>
150: #include <signal.h>
151: #include <sys/types.h>
152: #include <sys/stat.h>
153:
154: #define ARGVAL() (*++(*argv) || (--argc && *++argv))
155:
156: int n_bits; /* number of bits/code */
157: int maxbits = BITS; /* user settable max # bits/code */
158: code_int maxcode; /* maximum code, given n_bits */
159: code_int maxmaxcode = 1L << BITS; /* should NEVER generate this code */
160: #ifdef COMPATIBLE /* But wrong! */
161: # define MAXCODE(n_bits) (1L << (n_bits) - 1)
162: #else
163: # define MAXCODE(n_bits) ((1L << (n_bits)) - 1)
164: #endif /* COMPATIBLE */
165:
166: #ifdef XENIX_16
167: count_int htab0[8192];
168: count_int htab1[8192];
169: count_int htab2[8192];
170: count_int htab3[8192];
171: count_int htab4[8192];
172: count_int htab5[8192];
173: count_int htab6[8192];
174: count_int htab7[8192];
175: count_int htab8[HSIZE-65536];
176: count_int * htab[9] = {
177: htab0, htab1, htab2, htab3, htab4, htab5, htab6, htab7, htab8 };
178:
179: #define htabof(i) (htab[(i) >> 13][(i) & 0x1fff])
180: unsigned short code0tab[16384];
181: unsigned short code1tab[16384];
182: unsigned short code2tab[16384];
183: unsigned short code3tab[16384];
184: unsigned short code4tab[16384];
185: unsigned short * codetab[5] = {
186: code0tab, code1tab, code2tab, code3tab, code4tab };
187:
188: #define codetabof(i) (codetab[(i) >> 14][(i) & 0x3fff])
189:
190: #else /* Normal machine */
191: # ifdef sel
192: /* support gould base register problems */
193: /*NOBASE*/
194: count_int htab [HSIZE];
195: unsigned short codetab [HSIZE];
196: /*NOBASE*/
197: # else /* !gould */
198: count_int htab [HSIZE];
199: unsigned short codetab [HSIZE];
200: # endif /* !gould */
201: #define htabof(i) htab[i]
202: #define codetabof(i) codetab[i]
203: #endif /* !XENIX_16 */
204: code_int hsize = HSIZE; /* for dynamic table sizing */
205: count_int fsize;
206:
207: /*
208: * To save much memory, we overlay the table used by compress() with those
209: * used by decompress(). The tab_prefix table is the same size and type
210: * as the codetab. The tab_suffix table needs 2**BITS characters. We
211: * get this from the beginning of htab. The output stack uses the rest
212: * of htab, and contains characters. There is plenty of room for any
213: * possible stack (stack used to be 8000 characters).
214: */
215:
216: #define tab_prefixof(i) codetabof(i)
217: #ifdef XENIX_16
218: # define tab_suffixof(i) ((char_type *)htab[(i)>>15])[(i) & 0x7fff]
219: # define de_stack ((char_type *)(htab2))
220: #else /* Normal machine */
221: # define tab_suffixof(i) ((char_type *)(htab))[i]
222: # define de_stack ((char_type *)&tab_suffixof(1<<BITS))
223: #endif /* XENIX_16 */
224:
225: code_int free_ent = 0; /* first unused entry */
226: int exit_stat = 0;
227:
228: code_int getcode();
229:
230: Usage() {
231: #ifdef DEBUG
232: fprintf(stderr,"Usage: compress [-dDVfc] [-b maxbits] [file ...]\n");
233: }
234: int debug = 0;
235: #else
236: fprintf(stderr,"Usage: compress [-dfvcV] [-b maxbits] [file ...]\n");
237: }
238: #endif /* DEBUG */
239: int nomagic = 0; /* Use a 3-byte magic number header, unless old file */
240: int zcat_flg = 0; /* Write output on stdout, suppress messages */
241: int quiet = 1; /* don't tell me about compression */
242:
243: /*
244: * block compression parameters -- after all codes are used up,
245: * and compression rate changes, start over.
246: */
247: int block_compress = BLOCK_MASK;
248: int clear_flg = 0;
249: long int ratio = 0;
250: #define CHECK_GAP 10000 /* ratio check interval */
251: count_int checkpoint = CHECK_GAP;
252: /*
253: * the next two codes should not be changed lightly, as they must not
254: * lie within the contiguous general code space.
255: */
256: #define FIRST 257 /* first free entry */
257: #define CLEAR 256 /* table clear output code */
258:
259: int force = 0;
260: char ofname [100];
261: #ifdef DEBUG
262: int verbose = 0;
263: #endif /* DEBUG */
264: int (*bgnd_flag)();
265:
266: int do_decomp = 0;
267:
268: /*****************************************************************
269: * TAG( main )
270: *
271: * Algorithm from "A Technique for High Performance Data Compression",
272: * Terry A. Welch, IEEE Computer Vol 17, No 6 (June 1984), pp 8-19.
273: *
274: * Usage: compress [-dfvc] [-b bits] [file ...]
275: * Inputs:
276: * -d: If given, decompression is done instead.
277: *
278: * -c: Write output on stdout, don't remove original.
279: *
280: * -b: Parameter limits the max number of bits/code.
281: *
282: * -f: Forces output file to be generated, even if one already
283: * exists, and even if no space is saved by compressing.
284: * If -f is not used, the user will be prompted if stdin is
285: * a tty, otherwise, the output file will not be overwritten.
286: *
287: * -v: Write compression statistics
288: *
289: * file ...: Files to be compressed. If none specified, stdin
290: * is used.
291: * Outputs:
292: * file.Z: Compressed form of file with same mode, owner, and utimes
293: * or stdout (if stdin used as input)
294: *
295: * Assumptions:
296: * When filenames are given, replaces with the compressed version
297: * (.Z suffix) only if the file decreases in size.
298: * Algorithm:
299: * Modified Lempel-Ziv method (LZW). Basically finds common
300: * substrings and replaces them with a variable size code. This is
301: * deterministic, and can be done on the fly. Thus, the decompression
302: * procedure needs no input table, but tracks the way the table was built.
303: */
304:
305: main( argc, argv )
306: register int argc; char **argv;
307: {
308: int overwrite = 0; /* Do not overwrite unless given -f flag */
309: char tempname[100];
310: char **filelist, **fileptr;
311: char *cp, *rindex(), *malloc();
312: struct stat statbuf;
313: extern onintr(), oops();
314:
315:
316: if ( (bgnd_flag = signal ( SIGINT, SIG_IGN )) != SIG_IGN ) {
317: signal ( SIGINT, onintr );
318: signal ( SIGSEGV, oops );
319: }
320:
321: #ifdef COMPATIBLE
322: nomagic = 1; /* Original didn't have a magic number */
323: #endif /* COMPATIBLE */
324:
325: filelist = fileptr = (char **)(malloc(argc * sizeof(*argv)));
326: *filelist = NULL;
327:
328: if((cp = rindex(argv[0], '/')) != 0) {
329: cp++;
330: } else {
331: cp = argv[0];
332: }
333: if(strcmp(cp, "uncompress") == 0) {
334: do_decomp = 1;
335: } else if(strcmp(cp, "zcat") == 0) {
336: do_decomp = 1;
337: zcat_flg = 1;
338: }
339:
340: #ifdef BSD4_2
341: /* 4.2BSD dependent - take it out if not */
342: setlinebuf( stderr );
343: #endif /* BSD4_2 */
344:
345: /* Argument Processing
346: * All flags are optional.
347: * -D => debug
348: * -V => print Version; debug verbose
349: * -d => do_decomp
350: * -v => unquiet
351: * -f => force overwrite of output file
352: * -n => no header: useful to uncompress old files
353: * -b maxbits => maxbits. If -b is specified, then maxbits MUST be
354: * given also.
355: * -c => cat all output to stdout
356: * -C => generate output compatible with compress 2.0.
357: * if a string is left, must be an input filename.
358: */
359: for (argc--, argv++; argc > 0; argc--, argv++) {
360: if (**argv == '-') { /* A flag argument */
361: while (*++(*argv)) { /* Process all flags in this arg */
362: switch (**argv) {
363: #ifdef DEBUG
364: case 'D':
365: debug = 1;
366: break;
367: case 'V':
368: verbose = 1;
369: version();
370: break;
371: #else
372: case 'V':
373: version();
374: break;
375: #endif /* DEBUG */
376: case 'v':
377: quiet = 0;
378: break;
379: case 'd':
380: do_decomp = 1;
381: break;
382: case 'f':
383: case 'F':
384: overwrite = 1;
385: force = 1;
386: break;
387: case 'n':
388: nomagic = 1;
389: break;
390: case 'C':
391: block_compress = 0;
392: break;
393: case 'b':
394: if (!ARGVAL()) {
395: fprintf(stderr, "Missing maxbits\n");
396: Usage();
397: exit(1);
398: }
399: maxbits = atoi(*argv);
400: goto nextarg;
401: case 'c':
402: zcat_flg = 1;
403: break;
404: case 'q':
405: quiet = 1;
406: break;
407: #if COHERENT
408: case 'w':
409: if (!ARGVAL()) {
410: fprintf(stderr, "Missing filename for -w\n");
411: Usage();
412: exit(1);
413: }
414: goto nextarg;
415: #endif
416: default:
417: fprintf(stderr, "Unknown flag: '%c'; ", **argv);
418: Usage();
419: exit(1);
420: }
421: }
422: }
423: else { /* Input file name */
424: *fileptr++ = *argv; /* Build input file list */
425: *fileptr = NULL;
426: /* process nextarg; */
427: }
428: nextarg: continue;
429: }
430:
431: if(maxbits < INIT_BITS) maxbits = INIT_BITS;
432: if (maxbits > BITS) maxbits = BITS;
433: maxmaxcode = 1L << maxbits;
434:
435: if (*filelist != NULL) {
436: for (fileptr = filelist; *fileptr; fileptr++) {
437: exit_stat = 0;
438: if (do_decomp != 0) { /* DECOMPRESSION */
439: /* Check for .Z suffix */
440: if (strcmp(*fileptr + strlen(*fileptr) - 2, ".Z") != 0) {
441: /* No .Z: tack one on */
442: strcpy(tempname, *fileptr);
443: strcat(tempname, ".Z");
444: *fileptr = tempname;
445: }
446: /* Open input file */
447: if ((freopen(*fileptr, "r", stdin)) == NULL) {
448: perror(*fileptr); continue;
449: }
450: /* Check the magic number */
451: if (nomagic == 0) {
452: if ((getchar() != (magic_header[0] & 0xFF))
453: || (getchar() != (magic_header[1] & 0xFF))) {
454: fprintf(stderr, "%s: not in compressed format\n",
455: *fileptr);
456: continue;
457: }
458: maxbits = getchar(); /* set -b from file */
459: block_compress = maxbits & BLOCK_MASK;
460: maxbits &= BIT_MASK;
461: maxmaxcode = 1L << maxbits;
462: if(maxbits > BITS) {
463: fprintf(stderr,
464: "%s: compressed with %d bits, can only handle %d bits\n",
465: *fileptr, maxbits, BITS);
466: continue;
467: }
468: }
469: /* Generate output filename */
470: strcpy(ofname, *fileptr);
471: ofname[strlen(*fileptr) - 2] = '\0'; /* Strip off .Z */
472: } else { /* COMPRESSION */
473: if (strcmp(*fileptr + strlen(*fileptr) - 2, ".Z") == 0) {
474: fprintf(stderr, "%s: already has .Z suffix -- no change\n",
475: *fileptr);
476: continue;
477: }
478: /* Open input file */
479: if ((freopen(*fileptr, "r", stdin)) == NULL) {
480: perror(*fileptr); continue;
481: }
482: stat ( *fileptr, &statbuf );
483: fsize = (long) statbuf.st_size;
484: /*
485: * tune hash table size for small files -- ad hoc,
486: * but the sizes match earlier #defines, which
487: * serve as upper bounds on the number of output codes.
488: */
489: hsize = HSIZE;
490: if ( fsize < (1 << 12) )
491: hsize = min ( 5003, HSIZE );
492: else if ( fsize < (1 << 13) )
493: hsize = min ( 9001, HSIZE );
494: else if ( fsize < (1 << 14) )
495: hsize = min ( 18013, HSIZE );
496: else if ( fsize < (1 << 15) )
497: hsize = min ( 35023, HSIZE );
498: else if ( fsize < 47000 )
499: hsize = min ( 50021, HSIZE );
500:
501: /* Generate output filename */
502: strcpy(ofname, *fileptr);
503: #ifndef BSD4_2 /* Short filenames */
504: if ((cp=rindex(ofname,'/')) != NULL) cp++;
505: else cp = ofname;
506: if (strlen(cp) > 12) {
507: fprintf(stderr,"%s: filename too long to tack on .Z\n",cp);
508: continue;
509: }
510: #endif /* BSD4_2 Long filenames allowed */
511: strcat(ofname, ".Z");
512: }
513: /* Check for overwrite of existing file */
514: if (overwrite == 0 && zcat_flg == 0) {
515: if (stat(ofname, &statbuf) == 0) {
516: char response[2];
517: response[0] = 'n';
518: fprintf(stderr, "%s already exists;", ofname);
519: if (foreground()) {
520: fprintf(stderr, " do you wish to overwrite %s (y or n)? ",
521: ofname);
522: fflush(stderr);
523: read(2, response, 2);
524: while (response[1] != '\n') {
525: if (read(2, response+1, 1) < 0) { /* Ack! */
526: perror("stderr"); break;
527: }
528: }
529: }
530: if (response[0] != 'y') {
531: fprintf(stderr, "\tnot overwritten\n");
532: continue;
533: }
534: }
535: }
536: if(zcat_flg == 0) { /* Open output file */
537: if (freopen(ofname, "w", stdout) == NULL) {
538: perror(ofname);
539: continue;
540: }
541: if(!quiet)
542: fprintf(stderr, "%s: ", *fileptr);
543: }
544:
545: /* Actually do the compression/decompression */
546: if (do_decomp == 0) compress();
547: #ifndef DEBUG
548: else decompress();
549: #else
550: else if (debug == 0) decompress();
551: else printcodes();
552: if (verbose) dump_tab();
553: #endif /* DEBUG */
554: if(zcat_flg == 0) {
555: copystat(*fileptr, ofname); /* Copy stats */
556: if((exit_stat == 1) || (!quiet))
557: putc('\n', stderr);
558: }
559: }
560: } else { /* Standard input */
561: if (do_decomp == 0) {
562: compress();
563: #ifdef DEBUG
564: if(verbose) dump_tab();
565: #endif /* DEBUG */
566: if(!quiet)
567: putc('\n', stderr);
568: } else {
569: /* Check the magic number */
570: if (nomagic == 0) {
571: if ((getchar()!=(magic_header[0] & 0xFF))
572: || (getchar()!=(magic_header[1] & 0xFF))) {
573: fprintf(stderr, "stdin: not in compressed format\n");
574: exit(1);
575: }
576: maxbits = getchar(); /* set -b from file */
577: block_compress = maxbits & BLOCK_MASK;
578: maxbits &= BIT_MASK;
579: maxmaxcode = 1L << maxbits;
580: fsize = 100000; /* assume stdin large for USERMEM */
581: if(maxbits > BITS) {
582: fprintf(stderr,
583: "stdin: compressed with %d bits, can only handle %d bits\n",
584: maxbits, BITS);
585: exit(1);
586: }
587: }
588: #ifndef DEBUG
589: decompress();
590: #else
591: if (debug == 0) decompress();
592: else printcodes();
593: if (verbose) dump_tab();
594: #endif /* DEBUG */
595: }
596: }
597: exit(exit_stat);
598: }
599:
600: static int offset;
601: long int in_count = 1; /* length of input */
602: long int bytes_out; /* length of compressed output */
603: long int out_count = 0; /* # of codes output (for debugging) */
604:
605: /*
606: * compress stdin to stdout
607: *
608: * Algorithm: use open addressing double hashing (no chaining) on the
609: * prefix code / next character combination. We do a variant of Knuth's
610: * algorithm D (vol. 3, sec. 6.4) along with G. Knott's relatively-prime
611: * secondary probe. Here, the modular division first probe is gives way
612: * to a faster exclusive-or manipulation. Also do block compression with
613: * an adaptive reset, whereby the code table is cleared when the compression
614: * ratio decreases, but after the table fills. The variable-length output
615: * codes are re-sized at this point, and a special CLEAR code is generated
616: * for the decompressor. Late addition: construct the table according to
617: * file size for noticeable speed improvement on small files. Please direct
618: * questions about this implementation to ames!jaw.
619: */
620:
621: compress() {
622: register long fcode;
623: register code_int i = 0;
624: register int c;
625: register code_int ent;
626: #ifdef XENIX_16
627: register code_int disp;
628: #else /* Normal machine */
629: register int disp;
630: #endif
631: register code_int hsize_reg;
632: register int hshift;
633:
634: #ifndef COMPATIBLE
635: if (nomagic == 0) {
636: putchar(magic_header[0]); putchar(magic_header[1]);
637: putchar((char)(maxbits | block_compress));
638: if(ferror(stdout))
639: writeerr();
640: }
641: #endif /* COMPATIBLE */
642:
643: offset = 0;
644: bytes_out = 3; /* includes 3-byte header mojo */
645: out_count = 0;
646: clear_flg = 0;
647: ratio = 0;
648: in_count = 1;
649: checkpoint = CHECK_GAP;
650: maxcode = MAXCODE(n_bits = INIT_BITS);
651: free_ent = ((block_compress) ? FIRST : 256 );
652:
653: ent = getchar ();
654:
655: hshift = 0;
656: for ( fcode = (long) hsize; fcode < 65536L; fcode *= 2L )
657: hshift++;
658: hshift = 8 - hshift; /* set hash code range bound */
659:
660: hsize_reg = hsize;
661: cl_hash( (count_int) hsize_reg); /* clear hash table */
662:
663: #ifdef SIGNED_COMPARE_SLOW
664: while ( (c = getchar()) != (unsigned) EOF ) {
665: #else
666: while ( (c = getchar()) != EOF ) {
667: #endif
668: in_count++;
669: fcode = (long) (((long) c << maxbits) + ent);
670: i = (((long)c << hshift) ^ ent); /* xor hashing */
671:
672: if ( htabof (i) == fcode ) {
673: ent = codetabof (i);
674: continue;
675: } else if ( (long)htabof (i) < 0 ) /* empty slot */
676: goto nomatch;
677: disp = hsize_reg - i; /* secondary hash (after G. Knott) */
678: if ( i == 0 )
679: disp = 1;
680: probe:
681: if ( (i -= disp) < 0 )
682: i += hsize_reg;
683:
684: if ( htabof (i) == fcode ) {
685: ent = codetabof (i);
686: continue;
687: }
688: if ( (long)htabof (i) > 0 )
689: goto probe;
690: nomatch:
691: output ( (code_int) ent );
692: out_count++;
693: ent = c;
694: #ifdef SIGNED_COMPARE_SLOW
695: if ( (unsigned) free_ent < (unsigned) maxmaxcode) {
696: #else
697: if ( free_ent < maxmaxcode ) {
698: #endif
699: codetabof (i) = free_ent++; /* code -> hashtable */
700: htabof (i) = fcode;
701: }
702: else if ( (count_int)in_count >= checkpoint && block_compress )
703: cl_block ();
704: }
705: /*
706: * Put out the final code.
707: */
708: output( (code_int)ent );
709: out_count++;
710: output( (code_int)-1 );
711:
712: /*
713: * Print out stats on stderr
714: */
715: if(zcat_flg == 0 && !quiet) {
716: #ifdef DEBUG
717: fprintf( stderr,
718: "%ld chars in, %ld codes (%ld bytes) out, compression factor: ",
719: in_count, out_count, bytes_out );
720: prratio( stderr, in_count, bytes_out );
721: fprintf( stderr, "\n");
722: fprintf( stderr, "\tCompression as in compact: " );
723: prratio( stderr, in_count-bytes_out, in_count );
724: fprintf( stderr, "\n");
725: fprintf( stderr, "\tLargest code (of last block) was %d (%d bits)\n",
726: free_ent - 1, n_bits );
727: #else /* !DEBUG */
728: fprintf( stderr, "Compression: " );
729: prratio( stderr, in_count-bytes_out, in_count );
730: #endif /* DEBUG */
731: }
732: if(bytes_out > in_count) /* exit(2) if no savings */
733: exit_stat = 2;
734: return;
735: }
736:
737: /*****************************************************************
738: * TAG( output )
739: *
740: * Output the given code.
741: * Inputs:
742: * code: A n_bits-bit integer. If == -1, then EOF. This assumes
743: * that n_bits =< (long)wordsize - 1.
744: * Outputs:
745: * Outputs code to the file.
746: * Assumptions:
747: * Chars are 8 bits long.
748: * Algorithm:
749: * Maintain a BITS character long buffer (so that 8 codes will
750: * fit in it exactly). Use the VAX insv instruction to insert each
751: * code in turn. When the buffer fills up empty it and start over.
752: */
753:
754: static char buf[BITS];
755:
756: #ifndef vax
757: char_type lmask[9] = {0xff, 0xfe, 0xfc, 0xf8, 0xf0, 0xe0, 0xc0, 0x80, 0x00};
758: char_type rmask[9] = {0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff};
759: #endif /* vax */
760:
761: output( code )
762: code_int code;
763: {
764: #ifdef DEBUG
765: static int col = 0;
766: #endif /* DEBUG */
767:
768: /*
769: * On the VAX, it is important to have the register declarations
770: * in exactly the order given, or the asm will break.
771: */
772: register int r_off = offset, bits= n_bits;
773: register char * bp = buf;
774:
775: #ifdef DEBUG
776: if ( verbose )
777: fprintf( stderr, "%5d%c", code,
778: (col+=6) >= 74 ? (col = 0, '\n') : ' ' );
779: #endif /* DEBUG */
780: if ( code >= 0 ) {
781: #ifdef vax
782: /* VAX DEPENDENT!! Implementation on other machines is below.
783: *
784: * Translation: Insert BITS bits from the argument starting at
785: * offset bits from the beginning of buf.
786: */
787: 0; /* Work around for pcc -O bug with asm and if stmt */
788: asm( "insv 4(ap),r11,r10,(r9)" );
789: #else /* not a vax */
790: /*
791: * byte/bit numbering on the VAX is simulated by the following code
792: */
793: /*
794: * Get to the first byte.
795: */
796: bp += (r_off >> 3);
797: r_off &= 7;
798: /*
799: * Since code is always >= 8 bits, only need to mask the first
800: * hunk on the left.
801: */
802: *bp = (*bp & rmask[r_off]) | (code << r_off) & lmask[r_off];
803: bp++;
804: bits -= (8 - r_off);
805: code >>= 8 - r_off;
806: /* Get any 8 bit parts in the middle (<=1 for up to 16 bits). */
807: if ( bits >= 8 ) {
808: *bp++ = code;
809: code >>= 8;
810: bits -= 8;
811: }
812: /* Last bits. */
813: if(bits)
814: *bp = code;
815: #endif /* vax */
816: offset += n_bits;
817: if ( offset == (n_bits << 3) ) {
818: bp = buf;
819: bits = n_bits;
820: bytes_out += bits;
821: do
822: putchar(*bp++);
823: while(--bits);
824: offset = 0;
825: }
826:
827: /*
828: * If the next entry is going to be too big for the code size,
829: * then increase it, if possible.
830: */
831: if ( free_ent > maxcode || (clear_flg > 0))
832: {
833: /*
834: * Write the whole buffer, because the input side won't
835: * discover the size increase until after it has read it.
836: */
837: if ( offset > 0 ) {
838: if( fwrite( buf, 1, n_bits, stdout ) != n_bits)
839: writeerr();
840: bytes_out += n_bits;
841: }
842: offset = 0;
843:
844: if ( clear_flg ) {
845: maxcode = MAXCODE (n_bits = INIT_BITS);
846: clear_flg = 0;
847: }
848: else {
849: n_bits++;
850: if ( n_bits == maxbits )
851: maxcode = maxmaxcode;
852: else
853: maxcode = MAXCODE(n_bits);
854: }
855: #ifdef DEBUG
856: if ( debug ) {
857: fprintf( stderr, "\nChange to %d bits\n", n_bits );
858: col = 0;
859: }
860: #endif /* DEBUG */
861: }
862: } else {
863: /*
864: * At EOF, write the rest of the buffer.
865: */
866: if ( offset > 0 )
867: fwrite( buf, 1, (offset + 7) / 8, stdout );
868: bytes_out += (offset + 7) / 8;
869: offset = 0;
870: fflush( stdout );
871: #ifdef DEBUG
872: if ( verbose )
873: fprintf( stderr, "\n" );
874: #endif /* DEBUG */
875: if( ferror( stdout ) )
876: writeerr();
877: }
878: }
879:
880: /*
881: * Decompress stdin to stdout. This routine adapts to the codes in the
882: * file building the "string" table on-the-fly; requiring no table to
883: * be stored in the compressed file. The tables used herein are shared
884: * with those of the compress() routine. See the definitions above.
885: */
886:
887: decompress() {
888: register char_type *stackp;
889: register int finchar;
890: register code_int code, oldcode, incode;
891:
892: /*
893: * As above, initialize the first 256 entries in the table.
894: */
895: maxcode = MAXCODE(n_bits = INIT_BITS);
896: for ( code = 255; code >= 0; code-- ) {
897: tab_prefixof(code) = 0;
898: tab_suffixof(code) = (char_type)code;
899: }
900: free_ent = ((block_compress) ? FIRST : 256 );
901:
902: finchar = oldcode = getcode();
903: if(oldcode == -1) /* EOF already? */
904: return; /* Get out of here */
905: putchar( (char)finchar ); /* first code must be 8 bits = char */
906: if(ferror(stdout)) /* Crash if can't write */
907: writeerr();
908: stackp = de_stack;
909:
910: while ( (code = getcode()) > -1 ) {
911:
912: if ( (code == CLEAR) && block_compress ) {
913: for ( code = 255; code >= 0; code-- )
914: tab_prefixof(code) = 0;
915: clear_flg = 1;
916: free_ent = FIRST - 1;
917: if ( (code = getcode ()) == -1 ) /* O, untimely death! */
918: break;
919: }
920: incode = code;
921: /*
922: * Special case for KwKwK string.
923: */
924: if ( code >= free_ent ) {
925: *stackp++ = finchar;
926: code = oldcode;
927: }
928:
929: /*
930: * Generate output characters in reverse order
931: */
932: #ifdef SIGNED_COMPARE_SLOW
933: while ( ((unsigned long)code) >= ((unsigned long)256) ) {
934: #else
935: while ( code >= 256 ) {
936: #endif
937: *stackp++ = tab_suffixof(code);
938: code = tab_prefixof(code);
939: }
940: *stackp++ = finchar = tab_suffixof(code);
941:
942: /*
943: * And put them out in forward order
944: */
945: do
946: putchar ( *--stackp );
947: while ( stackp > de_stack );
948:
949: /*
950: * Generate the new entry.
951: */
952: if ( (code=free_ent) < maxmaxcode ) {
953: tab_prefixof(code) = (unsigned short)oldcode;
954: tab_suffixof(code) = finchar;
955: free_ent = code+1;
956: }
957: /*
958: * Remember previous code.
959: */
960: oldcode = incode;
961: }
962: fflush( stdout );
963: if(ferror(stdout))
964: writeerr();
965: }
966:
967: /*****************************************************************
968: * TAG( getcode )
969: *
970: * Read one code from the standard input. If EOF, return -1.
971: * Inputs:
972: * stdin
973: * Outputs:
974: * code or -1 is returned.
975: */
976:
977: code_int
978: getcode() {
979: /*
980: * On the VAX, it is important to have the register declarations
981: * in exactly the order given, or the asm will break.
982: */
983: register code_int code;
984: static int offset = 0, size = 0;
985: static char_type buf[BITS];
986: register int r_off, bits;
987: register char_type *bp = buf;
988:
989: if ( clear_flg > 0 || offset >= size || free_ent > maxcode ) {
990: /*
991: * If the next entry will be too big for the current code
992: * size, then we must increase the size. This implies reading
993: * a new buffer full, too.
994: */
995: if ( free_ent > maxcode ) {
996: n_bits++;
997: if ( n_bits == maxbits )
998: maxcode = maxmaxcode; /* won't get any bigger now */
999: else
1000: maxcode = MAXCODE(n_bits);
1001: }
1002: if ( clear_flg > 0) {
1003: maxcode = MAXCODE (n_bits = INIT_BITS);
1004: clear_flg = 0;
1005: }
1006: size = fread( buf, 1, n_bits, stdin );
1007: if ( size <= 0 )
1008: return -1; /* end of file */
1009: offset = 0;
1010: /* Round size down to integral number of codes */
1011: size = (size << 3) - (n_bits - 1);
1012: }
1013: r_off = offset;
1014: bits = n_bits;
1015: #ifdef vax
1016: asm( "extzv r10,r9,(r8),r11" );
1017: #else /* not a vax */
1018: /*
1019: * Get to the first byte.
1020: */
1021: bp += (r_off >> 3);
1022: r_off &= 7;
1023: /* Get first part (low order bits) */
1024: #ifdef NO_UCHAR
1025: code = ((*bp++ >> r_off) & rmask[8 - r_off]) & 0xff;
1026: #else
1027: code = (*bp++ >> r_off);
1028: #endif /* NO_UCHAR */
1029: bits -= (8 - r_off);
1030: r_off = 8 - r_off; /* now, offset into code word */
1031: /* Get any 8 bit parts in the middle (<=1 for up to 16 bits). */
1032: if ( bits >= 8 ) {
1033: #ifdef NO_UCHAR
1034: code |= (*bp++ & 0xff) << r_off;
1035: #else
1036: code |= *bp++ << r_off;
1037: #endif /* NO_UCHAR */
1038: r_off += 8;
1039: bits -= 8;
1040: }
1041: /* high order bits. */
1042: code |= (*bp & rmask[bits]) << r_off;
1043: #endif /* vax */
1044: offset += n_bits;
1045:
1046: return code;
1047: }
1048:
1049: char *
1050: rindex(s, c) /* For those who don't have it in libc.a */
1051: register char *s, c;
1052: {
1053: char *p;
1054: for (p = NULL; *s; s++)
1055: if (*s == c)
1056: p = s;
1057: return(p);
1058: }
1059:
1060: #ifdef DEBUG
1061: printcodes()
1062: {
1063: /*
1064: * Just print out codes from input file. For debugging.
1065: */
1066: code_int code;
1067: int col = 0, bits;
1068:
1069: bits = n_bits = INIT_BITS;
1070: maxcode = MAXCODE(n_bits);
1071: free_ent = ((block_compress) ? FIRST : 256 );
1072: while ( ( code = getcode() ) >= 0 ) {
1073: if ( (code == CLEAR) && block_compress ) {
1074: free_ent = FIRST - 1;
1075: clear_flg = 1;
1076: }
1077: else if ( free_ent < maxmaxcode )
1078: free_ent++;
1079: if ( bits != n_bits ) {
1080: fprintf(stderr, "\nChange to %d bits\n", n_bits );
1081: bits = n_bits;
1082: col = 0;
1083: }
1084: fprintf(stderr, "%5d%c", code, (col+=6) >= 74 ? (col = 0, '\n') : ' ' );
1085: }
1086: putc( '\n', stderr );
1087: exit( 0 );
1088: }
1089:
1090: #ifdef XENIX_16
1091: code_int stab1[8192] ;
1092: code_int stab2[8192] ;
1093: code_int stab3[8192] ;
1094: code_int stab4[8192] ;
1095: code_int stab5[8192] ;
1096: code_int stab6[8192] ;
1097: code_int stab7[8192] ;
1098: code_int stab8[8192] ;
1099: code_int * sorttab[8] = {stab1, stab2, stab3, stab4, stab5, stab6, stab7,
1100: stab8 } ;
1101: #define stabof(i) (sorttab[(i) >> 13][(i) & 0x1fff])
1102: #else
1103: code_int sorttab[HSIZE]; /* sorted pointers into htab */
1104: #define stabof(i) (sorttab[i])
1105: #endif
1106:
1107: dump_tab() /* dump string table */
1108: {
1109: register int i, first;
1110: register ent;
1111: #define STACK_SIZE 15000
1112: int stack_top = STACK_SIZE;
1113: register c;
1114: unsigned mbshift ;
1115:
1116: if(do_decomp == 0) { /* compressing */
1117: register int flag = 1;
1118:
1119: for(i=0; i<hsize; i++) { /* build sort pointers */
1120: if((long)htabof(i) >= 0) {
1121: stabof(codetabof(i)) = i;
1122: }
1123: }
1124: first = block_compress ? FIRST : 256;
1125: for(i = first; i < free_ent; i++) {
1126: fprintf(stderr, "%5d: \"", i);
1127: de_stack[--stack_top] = '\n';
1128: de_stack[--stack_top] = '"';
1129: stack_top = in_stack((htabof(stabof(i))>>maxbits)&0xff,
1130: stack_top);
1131: /* for(ent=htabof(stabof(i)) & ((1<<maxbits)-1); */
1132: mbshift = ((1 << maxbits) - 1) ;
1133: ent = htabof(stabof(i)) & mbshift ;
1134: for(;
1135: ent > 256;
1136: /* ent=htabof(stabof(ent)) & ((1<<maxbits)-1)) { */
1137: ent=htabof(stabof(ent)) & mbshift) {
1138: stack_top = in_stack(htabof(stabof(ent)) >> maxbits,
1139: stack_top);
1140: }
1141: stack_top = in_stack(ent, stack_top);
1142: fwrite( &de_stack[stack_top], 1, STACK_SIZE-stack_top, stderr);
1143: stack_top = STACK_SIZE;
1144: }
1145: } else if(!debug) { /* decompressing */
1146:
1147: for ( i = 0; i < free_ent; i++ ) {
1148: ent = i;
1149: c = tab_suffixof(ent);
1150: if ( isascii(c) && isprint(c) )
1151: fprintf( stderr, "%5d: %5d/'%c' \"",
1152: ent, tab_prefixof(ent), c );
1153: else
1154: fprintf( stderr, "%5d: %5d/\\%03o \"",
1155: ent, tab_prefixof(ent), c );
1156: de_stack[--stack_top] = '\n';
1157: de_stack[--stack_top] = '"';
1158: for ( ; ent != 0;
1159: ent = (ent >= FIRST ? tab_prefixof(ent) : 0) ) {
1160: stack_top = in_stack(tab_suffixof(ent), stack_top);
1161: }
1162: fwrite( &de_stack[stack_top], 1, STACK_SIZE - stack_top, stderr );
1163: stack_top = STACK_SIZE;
1164: }
1165: }
1166: }
1167:
1168: int
1169: in_stack(c, stack_top)
1170: register c, stack_top;
1171: {
1172: if ( (isascii(c) && isprint(c) && c != '\\') || c == ' ' ) {
1173: de_stack[--stack_top] = c;
1174: } else {
1175: switch( c ) {
1176: case '\n': de_stack[--stack_top] = 'n'; break;
1177: case '\t': de_stack[--stack_top] = 't'; break;
1178: case '\b': de_stack[--stack_top] = 'b'; break;
1179: case '\f': de_stack[--stack_top] = 'f'; break;
1180: case '\r': de_stack[--stack_top] = 'r'; break;
1181: case '\\': de_stack[--stack_top] = '\\'; break;
1182: default:
1183: de_stack[--stack_top] = '0' + c % 8;
1184: de_stack[--stack_top] = '0' + (c / 8) % 8;
1185: de_stack[--stack_top] = '0' + c / 64;
1186: break;
1187: }
1188: de_stack[--stack_top] = '\\';
1189: }
1190: return stack_top;
1191: }
1192: #endif /* DEBUG */
1193:
1194: writeerr()
1195: {
1196: perror ( ofname );
1197: unlink ( ofname );
1198: exit ( 1 );
1199: }
1200:
1201: copystat(ifname, ofname)
1202: char *ifname, *ofname;
1203: {
1204: struct stat statbuf;
1205: int mode;
1206: time_t timep[2];
1207:
1208: fclose(stdout);
1209: if (stat(ifname, &statbuf)) { /* Get stat on input file */
1210: perror(ifname);
1211: return;
1212: }
1213: if ((statbuf.st_mode & S_IFMT/*0170000*/) != S_IFREG/*0100000*/) {
1214: if(quiet)
1215: fprintf(stderr, "%s: ", ifname);
1216: fprintf(stderr, " -- not a regular file: unchanged");
1217: exit_stat = 1;
1218: } else if (statbuf.st_nlink > 1) {
1219: if(quiet)
1220: fprintf(stderr, "%s: ", ifname);
1221: fprintf(stderr, " -- has %d other links: unchanged",
1222: statbuf.st_nlink - 1);
1223: exit_stat = 1;
1224: } else if (exit_stat == 2 && (!force)) { /* No compression: remove file.Z */
1225: if(!quiet)
1226: fprintf(stderr, " -- file unchanged");
1227: } else { /* ***** Successful Compression ***** */
1228: exit_stat = 0;
1229: mode = statbuf.st_mode & 07777;
1230: if (chmod(ofname, mode)) /* Copy modes */
1231: perror(ofname);
1232: chown(ofname, statbuf.st_uid, statbuf.st_gid); /* Copy ownership */
1233: timep[0] = statbuf.st_atime;
1234: timep[1] = statbuf.st_mtime;
1235: utime(ofname, timep); /* Update last accessed and modified times */
1236: if (unlink(ifname)) /* Remove input file */
1237: perror(ifname);
1238: if(!quiet)
1239: fprintf(stderr, " -- replaced with %s", ofname);
1240: return; /* Successful return */
1241: }
1242:
1243: /* Unsuccessful return -- one of the tests failed */
1244: if (unlink(ofname))
1245: perror(ofname);
1246: }
1247: /*
1248: * This routine returns 1 if we are running in the foreground and stderr
1249: * is a tty.
1250: */
1251: foreground()
1252: {
1253: if(bgnd_flag) { /* background? */
1254: return(0);
1255: } else { /* foreground */
1256: if(isatty(2)) { /* and stderr is a tty */
1257: return(1);
1258: } else {
1259: return(0);
1260: }
1261: }
1262: }
1263:
1264: onintr ( )
1265: {
1266: unlink ( ofname );
1267: exit ( 1 );
1268: }
1269:
1270: oops ( ) /* wild pointer -- assume bad input */
1271: {
1272: if ( do_decomp == 1 )
1273: fprintf ( stderr, "uncompress: corrupt input\n" );
1274: unlink ( ofname );
1275: exit ( 1 );
1276: }
1277:
1278: cl_block () /* table clear for block compress */
1279: {
1280: register long int rat;
1281:
1282: checkpoint = in_count + CHECK_GAP;
1283: #ifdef DEBUG
1284: if ( debug ) {
1285: fprintf ( stderr, "count: %ld, ratio: ", in_count );
1286: prratio ( stderr, in_count, bytes_out );
1287: fprintf ( stderr, "\n");
1288: }
1289: #endif /* DEBUG */
1290:
1291: if(in_count > 0x007fffff) { /* shift will overflow */
1292: rat = bytes_out >> 8;
1293: if(rat == 0) { /* Don't divide by zero */
1294: rat = 0x7fffffff;
1295: } else {
1296: rat = in_count / rat;
1297: }
1298: } else {
1299: rat = (in_count << 8) / bytes_out; /* 8 fractional bits */
1300: }
1301: if ( rat > ratio ) {
1302: ratio = rat;
1303: } else {
1304: ratio = 0;
1305: #ifdef DEBUG
1306: if(verbose)
1307: dump_tab(); /* dump string table */
1308: #endif
1309: cl_hash ( (count_int) hsize );
1310: free_ent = FIRST;
1311: clear_flg = 1;
1312: output ( (code_int) CLEAR );
1313: #ifdef DEBUG
1314: if(debug)
1315: fprintf ( stderr, "clear\n" );
1316: #endif /* DEBUG */
1317: }
1318: }
1319:
1320: cl_hash(hsize) /* reset code table */
1321: register count_int hsize;
1322: {
1323: #ifndef XENIX_16 /* Normal machine */
1324: register count_int *htab_p = htab+hsize;
1325: #else
1326: register j;
1327: register long k = hsize;
1328: register count_int *htab_p;
1329: #endif
1330: register long i;
1331: register long m1 = -1;
1332:
1333: #ifdef XENIX_16
1334: for(j=0; j<=8 && k>=0; j++,k-=8192) {
1335: i = 8192;
1336: if(k < 8192) {
1337: i = k;
1338: }
1339: htab_p = &(htab[j][i]);
1340: i -= 16;
1341: if(i > 0) {
1342: #else
1343: i = hsize - 16;
1344: #endif
1345: do { /* might use Sys V memset(3) here */
1346: *(htab_p-16) = m1;
1347: *(htab_p-15) = m1;
1348: *(htab_p-14) = m1;
1349: *(htab_p-13) = m1;
1350: *(htab_p-12) = m1;
1351: *(htab_p-11) = m1;
1352: *(htab_p-10) = m1;
1353: *(htab_p-9) = m1;
1354: *(htab_p-8) = m1;
1355: *(htab_p-7) = m1;
1356: *(htab_p-6) = m1;
1357: *(htab_p-5) = m1;
1358: *(htab_p-4) = m1;
1359: *(htab_p-3) = m1;
1360: *(htab_p-2) = m1;
1361: *(htab_p-1) = m1;
1362: htab_p -= 16;
1363: } while ((i -= 16) >= 0);
1364: #ifdef XENIX_16
1365: }
1366: }
1367: #endif
1368: for ( i += 16; i > 0; i-- )
1369: *--htab_p = m1;
1370: }
1371:
1372: prratio(stream, num, den)
1373: FILE *stream;
1374: long int num, den;
1375: {
1376: register int q; /* Doesn't need to be long */
1377:
1378: if(num > 214748L) { /* 2147483647/10000 */
1379: q = num / (den / 10000L);
1380: } else {
1381: q = 10000L * num / den; /* Long calculations, though */
1382: }
1383: if (q < 0) {
1384: putc('-', stream);
1385: q = -q;
1386: }
1387: fprintf(stream, "%d.%02d%%", q / 100, q % 100);
1388: }
1389:
1390: version()
1391: {
1392: fprintf(stderr, "%s\n", rcs_ident);
1393: fprintf(stderr, "Options: ");
1394: #ifdef vax
1395: fprintf(stderr, "vax, ");
1396: #endif
1397: #ifdef NO_UCHAR
1398: fprintf(stderr, "NO_UCHAR, ");
1399: #endif
1400: #ifdef SIGNED_COMPARE_SLOW
1401: fprintf(stderr, "SIGNED_COMPARE_SLOW, ");
1402: #endif
1403: #ifdef XENIX_16
1404: fprintf(stderr, "XENIX_16, ");
1405: #endif
1406: #ifdef COMPATIBLE
1407: fprintf(stderr, "COMPATIBLE, ");
1408: #endif
1409: #ifdef DEBUG
1410: fprintf(stderr, "DEBUG, ");
1411: #endif
1412: #ifdef BSD4_2
1413: fprintf(stderr, "BSD4_2, ");
1414: #endif
1415: fprintf(stderr, "BITS = %d\n", BITS);
1416: }
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.