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         int fdesc;
 104
 105         while ((c = getopt(argc, argv, "-cv")) != -1)
 106                 switch (c) {
 107                 case '-':
 108                         readstd++;
 109                         break;
 110                 case 'c':
 111                         cflg++;
 112                         break;
 113                 case 'v':
 114                         vflg++;
 115                         break;
 116                 default:
 117                         usage();
 118                 }
 119         argc -= optind;
 120         argv += optind;
 121
 122         if (signal(SIGINT, SIG_IGN) == SIG_DFL)
 123                 signal(SIGINT, onintr);
 124         if (cflg || (argc == 0 && !readstd))
 125                 inithash();
 126         else {
 127                 strings = strdup(_PATH_TMP);
 128                 if (strings == NULL)
 129                         err(1, "strdup() failed");
 130                 fdesc = mkstemp(strings);
 131                 if (fdesc == -1)
 132                         err(1, "Unable to create temporary file");
 133                 close(fdesc);
 134         }
 135
 136         while (readstd || argc > 0) {
 137                 if (freopen("x.c", "w", stdout) == NULL)
 138                         err(1, "x.c");
 139                 if (!readstd && freopen(argv[0], "r", stdin) == NULL)
 140                         err(2, "%s", argv[0]);
 141                 process("x.c");
 142                 if (readstd == 0)
 143                         argc--, argv++;
 144                 else
 145                         readstd = 0;
 146         };
 147         flushsh();
 148         if (cflg == 0)
 149                 xsdotc();
 150         if (strings[0] == '/')
 151                 ignore(unlink(strings));
 152         exit(0);
 153 }
 154
 155 static void
 156 usage(void)
 157 {
 158         fprintf(stderr, "usage: xstr [-cv] [-] [file ...]\n");
 159         exit (1);
 160 }
 161
 162 char linebuf[BUFSIZ];
 163
 164 void
 165 process(const char *name)
 166 {
 167         char *cp;
 168         int c;
 169         int incomm = 0;
 170         int ret;
 171
 172         printf("extern char\txstr[];\n");
 173         for (;;) {
 174                 if (fgets(linebuf, sizeof linebuf, stdin) == NULL) {
 175                         if (ferror(stdin))
 176                                 err(3, "%s", name);
 177                         break;
 178                 }
 179                 if (linebuf[0] == '#') {
 180                         if (linebuf[1] == ' ' && isdigit(linebuf[2]))
 181                                 printf("#line%s", &linebuf[1]);
 182                         else
 183                                 printf("%s", linebuf);
 184                         continue;
 185                 }
 186                 for (cp = linebuf; (c = *cp++);) switch (c) {
 187
 188                 case '"':
 189                         if (incomm)
 190                                 goto def;
 191                         if ((ret = (int) yankstr(&cp)) == -1)
 192                                 goto out;
 193                         printf("(&xstr[%d])", ret);
 194                         break;
 195
 196                 case '\'':
 197                         if (incomm)
 198                                 goto def;
 199                         putchar(c);
 200                         if (*cp)
 201                                 putchar(*cp++);
 202                         break;
 203
 204                 case '/':
 205                         if (incomm || *cp != '*')
 206                                 goto def;
 207                         incomm = 1;
 208                         cp++;
 209                         printf("/*");
 210                         continue;
 211
 212                 case '*':
 213                         if (incomm && *cp == '/') {
 214                                 incomm = 0;
 215                                 cp++;
 216                                 printf("*/");
 217                                 continue;
 218                         }
 219                         goto def;
 220
 221 def:
 222                 default:
 223                         putchar(c);
 224                         break;
 225                 }
 226         }
 227 out:
 228         if (ferror(stdout))
 229                 warn("x.c"), onintr(0);
 230 }
 231
 232 off_t
 233 yankstr(char **cpp)
 234 {
 235         char *cp = *cpp;
 236         int c, ch;
 237         char dbuf[BUFSIZ];
 238         char *dp = dbuf;
 239         char *tp;
 240         static char tmp[] = "b\bt\tr\rn\nf\f\\\\\"\"";
 241
 242         while ((c = *cp++)) {
 243                 if (dp == dbuf + sizeof(dbuf) - 3)
 244                         errx(1, "message too long");
 245                 switch (c) {
 246
 247                 case '"':
 248                         cp++;
 249                         goto out;
 250
 251                 case '\\':
 252                         c = *cp++;
 253                         if (c == 0)
 254                                 break;
 255                         if (c == '\n') {
 256                                 if (fgets(linebuf, sizeof linebuf, stdin)
 257                                     == NULL) {
 258                                         if (ferror(stdin))
 259                                                 err(3, "x.c");
 260                                         return(-1);
 261                                 }
 262                                 cp = linebuf;
 263                                 continue;
 264                         }
 265                         for (tp = tmp; (ch = *tp++); tp++)
 266                                 if (c == ch) {
 267                                         c = *tp;
 268                                         goto gotc;
 269                                 }
 270                         if (!octdigit(c)) {
 271                                 *dp++ = '\\';
 272                                 break;
 273                         }
 274                         c -= '0';
 275                         if (!octdigit(*cp))
 276                                 break;
 277                         c <<= 3, c += *cp++ - '0';
 278                         if (!octdigit(*cp))
 279                                 break;
 280                         c <<= 3, c += *cp++ - '0';
 281                         break;
 282                 }
 283 gotc:
 284                 *dp++ = c;
 285         }
 286 out:
 287         *cpp = --cp;
 288         *dp = 0;
 289         return (hashit(dbuf, 1));
 290 }
 291
 292 int
 293 octdigit(char c)
 294 {
 295         return (isdigit(c) && c != '8' && c != '9');
 296 }
 297
 298 void
 299 inithash(void)
 300 {
 301         char buf[BUFSIZ];
 302         FILE *mesgread = fopen(strings, "r");
 303
 304         if (mesgread == NULL)
 305                 return;
 306         for (;;) {
 307                 mesgpt = tellpt;
 308                 if (fgetNUL(buf, sizeof buf, mesgread) == 0)
 309                         break;
 310                 ignore(hashit(buf, 0));
 311         }
 312         ignore(fclose(mesgread));
 313 }
 314
 315 int
 316 fgetNUL(char *obuf, int rmdr, FILE *file)
 317 {
 318         int c;
 319         char *buf = obuf;
 320
 321         while (--rmdr > 0 && (c = xgetc(file)) != 0 && c != EOF)
 322                 *buf++ = c;
 323         *buf++ = 0;
 324         return ((feof(file) || ferror(file)) ? 0 : 1);
 325 }
 326
 327 int
 328 xgetc(FILE *file)
 329 {
 330
 331         tellpt++;
 332         return (getc(file));
 333 }
 334
 335 #define BUCKETS 128
 336
 337 struct  hash {
 338         off_t   hpt;
 339         char    *hstr;
 340         struct  hash *hnext;
 341         short   hnew;
 342 } bucket[BUCKETS];
 343
 344 off_t
 345 hashit(char *str, int new)
 346 {
 347         int i;
 348         struct hash *hp, *hp0;
 349
 350         hp = hp0 = &bucket[lastchr(str) & 0177];
 351         while (hp->hnext) {
 352                 hp = hp->hnext;
 353                 i = istail(str, hp->hstr);
 354                 if (i >= 0)
 355                         return (hp->hpt + i);
 356         }
 357         if ((hp = (struct hash *) calloc(1, sizeof (*hp))) == NULL)
 358                 errx(8, "calloc");
 359         hp->hpt = mesgpt;
 360         if (!(hp->hstr = strdup(str)))
 361                 err(1, NULL);
 362         mesgpt += strlen(hp->hstr) + 1;
 363         hp->hnext = hp0->hnext;
 364         hp->hnew = new;
 365         hp0->hnext = hp;
 366         return (hp->hpt);
 367 }
 368
 369 void
 370 flushsh(void)
 371 {
 372         int i;
 373         struct hash *hp;
 374         FILE *mesgwrit;
 375         int old = 0, new = 0;
 376
 377         for (i = 0; i < BUCKETS; i++)
 378                 for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext)
 379                         if (hp->hnew)
 380                                 new++;
 381                         else
 382                                 old++;
 383         if (new == 0 && old != 0)
 384                 return;
 385         mesgwrit = fopen(strings, old ? "r+" : "w");
 386         if (mesgwrit == NULL)
 387                 err(4, "%s", strings);
 388         for (i = 0; i < BUCKETS; i++)
 389                 for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext) {
 390                         found(hp->hnew, hp->hpt, hp->hstr);
 391                         if (hp->hnew) {
 392                                 fseek(mesgwrit, hp->hpt, 0);
 393                                 ignore(fwrite(hp->hstr, strlen(hp->hstr) + 1, 1, mesgwrit));
 394                                 if (ferror(mesgwrit))
 395                                         err(4, "%s", strings);
 396                         }
 397                 }
 398         if (fclose(mesgwrit) == EOF)
 399                 err(4, "%s", strings);
 400 }
 401
 402 void
 403 found(int new, off_t off, char *str)
 404 {
 405         if (vflg == 0)
 406                 return;
 407         if (!new)
 408                 fprintf(stderr, "found at %d:", (int) off);
 409         else
 410                 fprintf(stderr, "new at %d:", (int) off);
 411         prstr(str);
 412         fprintf(stderr, "\n");
 413 }
 414
 415 void
 416 prstr(char *cp)
 417 {
 418         int c;
 419
 420         while ((c = (*cp++ & 0377)))
 421                 if (c < ' ')
 422                         fprintf(stderr, "^%c", c + '`');
 423                 else if (c == 0177)
 424                         fprintf(stderr, "^?");
 425                 else if (c > 0200)
 426                         fprintf(stderr, "\\%03o", c);
 427                 else
 428                         fprintf(stderr, "%c", c);
 429 }
 430
 431 void
 432 xsdotc(void)
 433 {
 434         FILE *strf = fopen(strings, "r");
 435         FILE *xdotcf;
 436
 437         if (strf == NULL)
 438                 err(5, "%s", strings);
 439         xdotcf = fopen("xs.c", "w");
 440         if (xdotcf == NULL)
 441                 err(6, "xs.c");
 442         fprintf(xdotcf, "char\txstr[] = {\n");
 443         for (;;) {
 444                 int i, c;
 445
 446                 for (i = 0; i < 8; i++) {
 447                         c = getc(strf);
 448                         if (ferror(strf)) {
 449                                 warn("%s", strings);
 450                                 onintr(0);
 451                         }
 452                         if (feof(strf)) {
 453                                 fprintf(xdotcf, "\n");
 454                                 goto out;
 455                         }
 456                         fprintf(xdotcf, "0x%02x,", c);
 457                 }
 458                 fprintf(xdotcf, "\n");
 459         }
 460 out:
 461         fprintf(xdotcf, "};\n");
 462         ignore(fclose(xdotcf));
 463         ignore(fclose(strf));
 464 }
 465
 466 char
 467 lastchr(char *cp)
 468 {
 469
 470         while (cp[0] && cp[1])
 471                 cp++;
 472         return (*cp);
 473 }
 474
 475 int
 476 istail(char *str, char *of)
 477 {
 478         int d = strlen(of) - strlen(str);
 479
 480         if (d < 0 || strcmp(&of[d], str) != 0)
 481                 return (-1);
 482         return (d);
 483 }
 484
 485 void
 486 onintr(int dummy __unused)
 487 {
 488
 489         ignore(signal(SIGINT, SIG_IGN));
 490         if (strings[0] == '/')
 491                 ignore(unlink(strings));
 492         ignore(unlink("x.c"));
 493         ignore(unlink("xs.c"));
 494         exit(7);
 495 }