usr.bin/xstr/xstr.c

   1 /*
   2  * Copyright (c) 1980, 1993
   3  *      The Regents of the University of California.  All rights reserved.
   4  *
   5  * Redistribution and use in source and binary forms, with or without
   6  * modification, are permitted provided that the following conditions
   7  * are met:
   8  * 1. Redistributions of source code must retain the above copyright
   9  *    notice, this list of conditions and the following disclaimer.
  10  * 2. Redistributions in binary form must reproduce the above copyright
  11  *    notice, this list of conditions and the following disclaimer in the
  12  *    documentation and/or other materials provided with the distribution.
  13  * 3. All advertising materials mentioning features or use of this software
  14  *    must display the following acknowledgement:
  15  *      This product includes software developed by the University of
  16  *      California, Berkeley and its contributors.
  17  * 4. Neither the name of the University nor the names of its contributors
  18  *    may be used to endorse or promote products derived from this software
  19  *    without specific prior written permission.
  20  *
  21  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  31  * SUCH DAMAGE.
  32  */
  33
  34 #include <sys/cdefs.h>
  35
  36 __FBSDID("$FreeBSD$");
  37
  38 #ifndef lint
  39 static const char copyright[] =
  40 "@(#) Copyright (c) 1980, 1993\n\
  41         The Regents of the University of California.  All rights reserved.\n";
  42 #endif
  43
  44 #ifndef lint
  45 static const char sccsid[] = "@(#)xstr.c        8.1 (Berkeley) 6/9/93";
  46 #endif
  47
  48 #include <sys/types.h>
  49
  50 #include <ctype.h>
  51 #include <err.h>
  52 #include <stdio.h>
  53 #include <stdlib.h>
  54 #include <signal.h>
  55 #include <string.h>
  56 #include <unistd.h>
  57
  58 #include "pathnames.h"
  59
  60 /*
  61  * xstr - extract and hash strings in a C program
  62  *
  63  * Bill Joy UCB
  64  * November, 1978
  65  */
  66
  67 #define ignore(a)       ((void) a)
  68
  69 off_t   tellpt;
  70
  71 off_t   mesgpt;
  72 char    cstrings[] =    "strings";
  73 char    *strings =      cstrings;
  74
  75 int     cflg;
  76 int     vflg;
  77 int     readstd;
  78
  79 char lastchr(char *);
  80
  81 int fgetNUL(char *, int, FILE *);
  82 int istail(char *, char *);
  83 int octdigit(char);
  84 int xgetc(FILE *);
  85
  86 off_t hashit(char *, int);
  87 off_t yankstr(char **);
  88
  89 static void usage(void);
  90
  91 void flushsh(void);
  92 void found(int, off_t, char *);
  93 void inithash(void);
  94 void onintr(int);
  95 void process(const char *);
  96 void prstr(char *);
  97 void xsdotc(void);
  98
  99 int
 100 main(int argc, char *argv[])
 101 {
 102         int c;
 103
 104         while ((c = getopt(argc, argv, "-cv")) != -1)
 105                 switch (c) {
 106                 case '-':
 107                         readstd++;
 108                         break;
 109                 case 'c':
 110                         cflg++;
 111                         break;
 112                 case 'v':
 113                         vflg++;
 114                         break;
 115                 default:
 116                         usage();
 117                 }
 118         argc -= optind;
 119         argv += optind;
 120
 121         if (signal(SIGINT, SIG_IGN) == SIG_DFL)
 122                 signal(SIGINT, onintr);
 123         if (cflg || (argc == 0 && !readstd))
 124                 inithash();
 125         else
 126                 strings = mktemp(strdup(_PATH_TMP));
 127         while (readstd || argc > 0) {
 128                 if (freopen("x.c", "w", stdout) == NULL)
 129                         err(1, "x.c");
 130                 if (!readstd && freopen(argv[0], "r", stdin) == NULL)
 131                         err(2, "%s", argv[0]);
 132                 process("x.c");
 133                 if (readstd == 0)
 134                         argc--, argv++;
 135                 else
 136                         readstd = 0;
 137         };
 138         flushsh();
 139         if (cflg == 0)
 140                 xsdotc();
 141         if (strings[0] == '/')
 142                 ignore(unlink(strings));
 143         exit(0);
 144 }
 145
 146 static void
 147 usage(void)
 148 {
 149         fprintf(stderr, "usage: xstr [-v] [-c] [-] [name ...]\n");
 150         exit (1);
 151 }
 152
 153 char linebuf[BUFSIZ];
 154
 155 void
 156 process(const char *name)
 157 {
 158         char *cp;
 159         int c;
 160         int incomm = 0;
 161         int ret;
 162
 163         printf("extern char\txstr[];\n");
 164         for (;;) {
 165                 if (fgets(linebuf, sizeof linebuf, stdin) == NULL) {
 166                         if (ferror(stdin))
 167                                 err(3, "%s", name);
 168                         break;
 169                 }
 170                 if (linebuf[0] == '#') {
 171                         if (linebuf[1] == ' ' && isdigit(linebuf[2]))
 172                                 printf("#line%s", &linebuf[1]);
 173                         else
 174                                 printf("%s", linebuf);
 175                         continue;
 176                 }
 177                 for (cp = linebuf; (c = *cp++);) switch (c) {
 178
 179                 case '"':
 180                         if (incomm)
 181                                 goto def;
 182                         if ((ret = (int) yankstr(&cp)) == -1)
 183                                 goto out;
 184                         printf("(&xstr[%d])", ret);
 185                         break;
 186
 187                 case '\'':
 188                         if (incomm)
 189                                 goto def;
 190                         putchar(c);
 191                         if (*cp)
 192                                 putchar(*cp++);
 193                         break;
 194
 195                 case '/':
 196                         if (incomm || *cp != '*')
 197                                 goto def;
 198                         incomm = 1;
 199                         cp++;
 200                         printf("/*");
 201                         continue;
 202
 203                 case '*':
 204                         if (incomm && *cp == '/') {
 205                                 incomm = 0;
 206                                 cp++;
 207                                 printf("*/");
 208                                 continue;
 209                         }
 210                         goto def;
 211
 212 def:
 213                 default:
 214                         putchar(c);
 215                         break;
 216                 }
 217         }
 218 out:
 219         if (ferror(stdout))
 220                 warn("x.c"), onintr(0);
 221 }
 222
 223 off_t
 224 yankstr(char **cpp)
 225 {
 226         char *cp = *cpp;
 227         int c, ch;
 228         char dbuf[BUFSIZ];
 229         char *dp = dbuf;
 230         char *tp;
 231         static char tmp[] = "b\bt\tr\rn\nf\f\\\\\"\"";
 232
 233         while ((c = *cp++)) {
 234                 switch (c) {
 235
 236                 case '"':
 237                         cp++;
 238                         goto out;
 239
 240                 case '\\':
 241                         c = *cp++;
 242                         if (c == 0)
 243                                 break;
 244                         if (c == '\n') {
 245                                 if (fgets(linebuf, sizeof linebuf, stdin)
 246                                     == NULL) {
 247                                         if (ferror(stdin))
 248                                                 err(3, "x.c");
 249                                         return(-1);
 250                                 }
 251                                 cp = linebuf;
 252                                 continue;
 253                         }
 254                         for (tp = tmp; (ch = *tp++); tp++)
 255                                 if (c == ch) {
 256                                         c = *tp;
 257                                         goto gotc;
 258                                 }
 259                         if (!octdigit(c)) {
 260                                 *dp++ = '\\';
 261                                 break;
 262                         }
 263                         c -= '0';
 264                         if (!octdigit(*cp))
 265                                 break;
 266                         c <<= 3, c += *cp++ - '0';
 267                         if (!octdigit(*cp))
 268                                 break;
 269                         c <<= 3, c += *cp++ - '0';
 270                         break;
 271                 }
 272 gotc:
 273                 *dp++ = c;
 274         }
 275 out:
 276         *cpp = --cp;
 277         *dp = 0;
 278         return (hashit(dbuf, 1));
 279 }
 280
 281 int
 282 octdigit(char c)
 283 {
 284         return (isdigit(c) && c != '8' && c != '9');
 285 }
 286
 287 void
 288 inithash(void)
 289 {
 290         char buf[BUFSIZ];
 291         FILE *mesgread = fopen(strings, "r");
 292
 293         if (mesgread == NULL)
 294                 return;
 295         for (;;) {
 296                 mesgpt = tellpt;
 297                 if (fgetNUL(buf, sizeof buf, mesgread) == 0)
 298                         break;
 299                 ignore(hashit(buf, 0));
 300         }
 301         ignore(fclose(mesgread));
 302 }
 303
 304 int
 305 fgetNUL(char *obuf, int rmdr, FILE *file)
 306 {
 307         int c;
 308         char *buf = obuf;
 309
 310         while (--rmdr > 0 && (c = xgetc(file)) != 0 && c != EOF)
 311                 *buf++ = c;
 312         *buf++ = 0;
 313         return ((feof(file) || ferror(file)) ? 0 : 1);
 314 }
 315
 316 int
 317 xgetc(FILE *file)
 318 {
 319
 320         tellpt++;
 321         return (getc(file));
 322 }
 323
 324 #define BUCKETS 128
 325
 326 struct  hash {
 327         off_t   hpt;
 328         char    *hstr;
 329         struct  hash *hnext;
 330         short   hnew;
 331 } bucket[BUCKETS];
 332
 333 off_t
 334 hashit(char *str, int new)
 335 {
 336         int i;
 337         struct hash *hp, *hp0;
 338
 339         hp = hp0 = &bucket[lastchr(str) & 0177];
 340         while (hp->hnext) {
 341                 hp = hp->hnext;
 342                 i = istail(str, hp->hstr);
 343                 if (i >= 0)
 344                         return (hp->hpt + i);
 345         }
 346         if ((hp = (struct hash *) calloc(1, sizeof (*hp))) == NULL)
 347                 errx(8, "calloc");
 348         hp->hpt = mesgpt;
 349         if (!(hp->hstr = strdup(str)))
 350                 err(1, NULL);
 351         mesgpt += strlen(hp->hstr) + 1;
 352         hp->hnext = hp0->hnext;
 353         hp->hnew = new;
 354         hp0->hnext = hp;
 355         return (hp->hpt);
 356 }
 357
 358 void
 359 flushsh(void)
 360 {
 361         int i;
 362         struct hash *hp;
 363         FILE *mesgwrit;
 364         int old = 0, new = 0;
 365
 366         for (i = 0; i < BUCKETS; i++)
 367                 for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext)
 368                         if (hp->hnew)
 369                                 new++;
 370                         else
 371                                 old++;
 372         if (new == 0 && old != 0)
 373                 return;
 374         mesgwrit = fopen(strings, old ? "r+" : "w");
 375         if (mesgwrit == NULL)
 376                 perror(strings), exit(4);
 377         for (i = 0; i < BUCKETS; i++)
 378                 for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext) {
 379                         found(hp->hnew, hp->hpt, hp->hstr);
 380                         if (hp->hnew) {
 381                                 fseek(mesgwrit, hp->hpt, 0);
 382                                 ignore(fwrite(hp->hstr, strlen(hp->hstr) + 1, 1, mesgwrit));
 383                                 if (ferror(mesgwrit))
 384                                         err(4, "%s", strings);
 385                         }
 386                 }
 387         if (fclose(mesgwrit) == EOF)
 388                 err(4, "%s", strings);
 389 }
 390
 391 void
 392 found(int new, off_t off, char *str)
 393 {
 394         if (vflg == 0)
 395                 return;
 396         if (!new)
 397                 fprintf(stderr, "found at %d:", (int) off);
 398         else
 399                 fprintf(stderr, "new at %d:", (int) off);
 400         prstr(str);
 401         fprintf(stderr, "\n");
 402 }
 403
 404 void
 405 prstr(char *cp)
 406 {
 407         int c;
 408
 409         while ((c = (*cp++ & 0377)))
 410                 if (c < ' ')
 411                         fprintf(stderr, "^%c", c + '`');
 412                 else if (c == 0177)
 413                         fprintf(stderr, "^?");
 414                 else if (c > 0200)
 415                         fprintf(stderr, "\\%03o", c);
 416                 else
 417                         fprintf(stderr, "%c", c);
 418 }
 419
 420 void
 421 xsdotc(void)
 422 {
 423         FILE *strf = fopen(strings, "r");
 424         FILE *xdotcf;
 425
 426         if (strf == NULL)
 427                 err(5, "%s", strings);
 428         xdotcf = fopen("xs.c", "w");
 429         if (xdotcf == NULL)
 430                 err(6, "xs.c");
 431         fprintf(xdotcf, "char\txstr[] = {\n");
 432         for (;;) {
 433                 int i, c;
 434
 435                 for (i = 0; i < 8; i++) {
 436                         c = getc(strf);
 437                         if (ferror(strf)) {
 438                                 warn("%s", strings);
 439                                 onintr(0);
 440                         }
 441                         if (feof(strf)) {
 442                                 fprintf(xdotcf, "\n");
 443                                 goto out;
 444                         }
 445                         fprintf(xdotcf, "0x%02x,", c);
 446                 }
 447                 fprintf(xdotcf, "\n");
 448         }
 449 out:
 450         fprintf(xdotcf, "};\n");
 451         ignore(fclose(xdotcf));
 452         ignore(fclose(strf));
 453 }
 454
 455 char
 456 lastchr(char *cp)
 457 {
 458
 459         while (cp[0] && cp[1])
 460                 cp++;
 461         return (*cp);
 462 }
 463
 464 int
 465 istail(char *str, char *of)
 466 {
 467         int d = strlen(of) - strlen(str);
 468
 469         if (d < 0 || strcmp(&of[d], str) != 0)
 470                 return (-1);
 471         return (d);
 472 }
 473
 474 void
 475 onintr(int dummy __unused)
 476 {
 477
 478         ignore(signal(SIGINT, SIG_IGN));
 479         if (strings[0] == '/')
 480                 ignore(unlink(strings));
 481         ignore(unlink("x.c"));
 482         ignore(unlink("xs.c"));
 483         exit(7);
 484 }