1 /* 2 * Copyright (c) 1989, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * This code is derived from software contributed to Berkeley by 6 * James A. Woods. 7 * 8 * %sccs.include.redist.c% 9 */ 10 11 #ifndef lint 12 static char copyright[] = 13 "@(#) Copyright (c) 1989, 1993\n\ 14 The Regents of the University of California. All rights reserved.\n"; 15 #endif /* not lint */ 16 17 #ifndef lint 18 static char sccsid[] = "@(#)locate.code.c 8.4 (Berkeley) 05/04/95"; 19 #endif /* not lint */ 20 21 /* 22 * PURPOSE: sorted list compressor (works with a modified 'find' 23 * to encode/decode a filename database) 24 * 25 * USAGE: bigram < list > bigrams 26 * process bigrams (see updatedb) > common_bigrams 27 * code common_bigrams < list > squozen_list 28 * 29 * METHOD: Uses 'front compression' (see ";login:", Volume 8, Number 1 30 * February/March 1983, p. 8). Output format is, per line, an 31 * offset differential count byte followed by a partially bigram- 32 * encoded ascii residue. A bigram is a two-character sequence, 33 * the first 128 most common of which are encoded in one byte. 34 * 35 * EXAMPLE: For simple front compression with no bigram encoding, 36 * if the input is... then the output is... 37 * 38 * /usr/src 0 /usr/src 39 * /usr/src/cmd/aardvark.c 8 /cmd/aardvark.c 40 * /usr/src/cmd/armadillo.c 14 armadillo.c 41 * /usr/tmp/zoo 5 tmp/zoo 42 * 43 * The codes are: 44 * 45 * 0-28 likeliest differential counts + offset to make nonnegative 46 * 30 switch code for out-of-range count to follow in next word 47 * 128-255 bigram codes (128 most common, as determined by 'updatedb') 48 * 32-127 single character (printable) ascii residue (ie, literal) 49 * 50 * SEE ALSO: updatedb.csh, bigram.c 51 * 52 * AUTHOR: James A. Woods, Informatics General Corp., 53 * NASA Ames Research Center, 10/82 54 */ 55 56 #include <sys/param.h> 57 58 #include <err.h> 59 #include <errno.h> 60 #include <stdio.h> 61 #include <stdlib.h> 62 #include <string.h> 63 #include <unistd.h> 64 65 #include "locate.h" 66 67 #define BGBUFSIZE (NBG * 2) /* size of bigram buffer */ 68 69 char buf1[MAXPATHLEN] = " "; 70 char buf2[MAXPATHLEN]; 71 char bigrams[BGBUFSIZE + 1] = { 0 }; 72 73 int bgindex __P((char *)); 74 void usage __P((void)); 75 76 int 77 main(argc, argv) 78 int argc; 79 char *argv[]; 80 { 81 register char *cp, *oldpath, *path; 82 int ch, code, count, diffcount, oldcount; 83 FILE *fp; 84 85 while ((ch = getopt(argc, argv, "")) != EOF) 86 switch(ch) { 87 case '?': 88 default: 89 usage(); 90 } 91 argc -= optind; 92 argv += optind; 93 94 if (argc != 1) 95 usage(); 96 97 if ((fp = fopen(argv[0], "r")) == NULL) 98 err(1, "%s", argv[0]); 99 100 /* First copy bigram array to stdout. */ 101 (void)fgets(bigrams, BGBUFSIZE + 1, fp); 102 if (fwrite(bigrams, 1, BGBUFSIZE, stdout) != BGBUFSIZE) 103 err(1, "stdout"); 104 (void)fclose(fp); 105 106 oldpath = buf1; 107 path = buf2; 108 oldcount = 0; 109 while (fgets(path, sizeof(buf2), stdin) != NULL) { 110 /* Truncate newline. */ 111 cp = path + strlen(path) - 1; 112 if (cp > path && *cp == '\n') 113 *cp = '\0'; 114 115 /* Squelch characters that would botch the decoding. */ 116 for (cp = path; *cp != '\0'; cp++) { 117 if ((u_char)*cp >= PARITY || *cp <= SWITCH) 118 *cp = '?'; 119 } 120 121 /* Skip longest common prefix. */ 122 for (cp = path; *cp == *oldpath; cp++, oldpath++) 123 if (*oldpath == '\0') 124 break; 125 count = cp - path; 126 diffcount = count - oldcount + OFFSET; 127 oldcount = count; 128 if (diffcount < 0 || diffcount > 2 * OFFSET) { 129 if (putchar(SWITCH) == EOF || 130 putw(diffcount, stdout) == EOF) 131 err(1, "stdout"); 132 } else 133 if (putchar(diffcount) == EOF) 134 err(1, "stdout"); 135 136 while (*cp != '\0') { 137 if (*(cp + 1) == '\0') { 138 if (putchar(*cp) == EOF) 139 err(1, "stdout"); 140 break; 141 } 142 if ((code = bgindex(cp)) < 0) { 143 if (putchar(*cp++) == EOF || 144 putchar(*cp++) == EOF) 145 err(1, "stdout"); 146 } else { 147 /* Found, so mark byte with parity bit. */ 148 if (putchar((code / 2) | PARITY) == EOF) 149 err(1, "stdout"); 150 cp += 2; 151 } 152 } 153 if (path == buf1) { /* swap pointers */ 154 path = buf2; 155 oldpath = buf1; 156 } else { 157 path = buf1; 158 oldpath = buf2; 159 } 160 } 161 /* Non-zero status if there were errors */ 162 if (fflush(stdout) != 0 || ferror(stdout)) 163 exit(1); 164 exit(0); 165 } 166 167 int 168 bgindex(bg) /* Return location of bg in bigrams or -1. */ 169 char *bg; 170 { 171 register char bg0, bg1, *p; 172 173 bg0 = bg[0]; 174 bg1 = bg[1]; 175 for (p = bigrams; *p != '\0'; p++) 176 if (*p++ == bg0 && *p == bg1) 177 break; 178 return (*p == '\0' ? -1 : --p - bigrams); 179 } 180 181 void 182 usage() 183 { 184 (void)fprintf(stderr, 185 "usage: locate.code common_bigrams < list > squozen_list\n"); 186 exit(1); 187 } 188