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. Neither the name of the University nor the names of its contributors
  14  *    may be used to endorse or promote products derived from this software
  15  *    without specific prior written permission.
  16  *
  17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  27  * SUCH DAMAGE.
  28  *
  29  * @(#) Copyright (c) 1980, 1993 The Regents of the University of California.  All rights reserved.
  30  * @(#)xstr.c   8.1 (Berkeley) 6/9/93
  31  * $FreeBSD: src/usr.bin/xstr/xstr.c,v 1.11 2008/05/13 09:42:03 kevlo Exp $
  32  */
  33
  34 #include <sys/types.h>
  35
  36 #include <ctype.h>
  37 #include <err.h>
  38 #include <stdio.h>
  39 #include <stdlib.h>
  40 #include <signal.h>
  41 #include <string.h>
  42 #include <unistd.h>
  43
  44 #include "pathnames.h"
  45
  46 /*
  47  * xstr - extract and hash strings in a C program
  48  *
  49  * Bill Joy UCB
  50  * November, 1978
  51  */
  52
  53 #define ignore(a)       ((void) a)
  54
  55 off_t   tellpt;
  56
  57 off_t   mesgpt;
  58 char    cstrings[] =    "strings";
  59 char    *strings =      cstrings;
  60
  61 int     cflg;
  62 int     vflg;
  63 int     readstd;
  64
  65 char lastchr(char *);
  66
  67 int fgetNUL(char *, int, FILE *);
  68 int istail(char *, char *);
  69 int octdigit(char);
  70 int xgetc(FILE *);
  71
  72 off_t hashit(char *, int);
  73 off_t yankstr(char **);
  74
  75 static void usage(void);
  76
  77 void flushsh(void);
  78 void found(int, off_t, char *);
  79 void inithash(void);
  80 void onintr(int);
  81 void process(const char *);
  82 void prstr(char *);
  83 void xsdotc(void);
  84
  85 int
  86 main(int argc, char *argv[])
  87 {
  88         int c;
  89         int fdesc;
  90
  91         while ((c = getopt(argc, argv, "-cv")) != -1)
  92                 switch (c) {
  93                 case '-':
  94                         readstd++;
  95                         break;
  96                 case 'c':
  97                         cflg++;
  98                         break;
  99                 case 'v':
 100                         vflg++;
 101                         break;
 102                 default:
 103                         usage();
 104                 }
 105         argc -= optind;
 106         argv += optind;
 107
 108         if (signal(SIGINT, SIG_IGN) == SIG_DFL)
 109                 signal(SIGINT, onintr);
 110         if (cflg || (argc == 0 && !readstd))
 111                 inithash();
 112         else {
 113                 strings = strdup(_PATH_TMP);
 114                 if (strings == NULL)
 115                         err(1, "strdup() failed");
 116                 fdesc = mkstemp(strings);
 117                 if (fdesc == -1)
 118                         err(1, "Unable to create temporary file");
 119                 close(fdesc);
 120         }
 121
 122         while (readstd || argc > 0) {
 123                 if (freopen("x.c", "w", stdout) == NULL)
 124                         err(1, "x.c");
 125                 if (!readstd && freopen(argv[0], "r", stdin) == NULL)
 126                         err(2, "%s", argv[0]);
 127                 process("x.c");
 128                 if (readstd == 0)
 129                         argc--, argv++;
 130                 else
 131                         readstd = 0;
 132         }
 133         flushsh();
 134         if (cflg == 0)
 135                 xsdotc();
 136         if (strings[0] == '/')
 137                 ignore(unlink(strings));
 138         exit(0);
 139 }
 140
 141 static void
 142 usage(void)
 143 {
 144         fprintf(stderr, "usage: xstr [-cv] [-] [file ...]\n");
 145         exit (1);
 146 }
 147
 148 char linebuf[BUFSIZ];
 149
 150 void
 151 process(const char *name)
 152 {
 153         char *cp;
 154         int c;
 155         int incomm = 0;
 156         int ret;
 157
 158         printf("extern char\txstr[];\n");
 159         for (;;) {
 160                 if (fgets(linebuf, sizeof linebuf, stdin) == NULL) {
 161                         if (ferror(stdin))
 162                                 err(3, "%s", name);
 163                         break;
 164                 }
 165                 if (linebuf[0] == '#') {
 166                         if (linebuf[1] == ' ' && isdigit(linebuf[2]))
 167                                 printf("#line%s", &linebuf[1]);
 168                         else
 169                                 printf("%s", linebuf);
 170                         continue;
 171                 }
 172                 for (cp = linebuf; (c = *cp++);) switch (c) {
 173
 174                 case '"':
 175                         if (incomm)
 176                                 goto def;
 177                         if ((ret = (int) yankstr(&cp)) == -1)
 178                                 goto out;
 179                         printf("(&xstr[%d])", ret);
 180                         break;
 181
 182                 case '\'':
 183                         if (incomm)
 184                                 goto def;
 185                         putchar(c);
 186                         if (*cp)
 187                                 putchar(*cp++);
 188                         break;
 189
 190                 case '/':
 191                         if (incomm || *cp != '*')
 192                                 goto def;
 193                         incomm = 1;
 194                         cp++;
 195                         printf("/*");
 196                         continue;
 197
 198                 case '*':
 199                         if (incomm && *cp == '/') {
 200                                 incomm = 0;
 201                                 cp++;
 202                                 printf("*/");
 203                                 continue;
 204                         }
 205                         goto def;
 206
 207 def:
 208                 default:
 209                         putchar(c);
 210                         break;
 211                 }
 212         }
 213 out:
 214         if (ferror(stdout))
 215                 warn("x.c"), onintr(0);
 216 }
 217
 218 off_t
 219 yankstr(char **cpp)
 220 {
 221         char *cp = *cpp;
 222         int c, ch;
 223         char dbuf[BUFSIZ];
 224         char *dp = dbuf;
 225         char *tp;
 226         static char tmp[] = "b\bt\tr\rn\nf\f\\\\\"\"";
 227
 228         while ((c = *cp++)) {
 229                 if (dp == dbuf + sizeof(dbuf) - 3)
 230                         errx(1, "message too long");
 231                 switch (c) {
 232
 233                 case '"':
 234                         cp++;
 235                         goto out;
 236
 237                 case '\\':
 238                         c = *cp++;
 239                         if (c == 0)
 240                                 break;
 241                         if (c == '\n') {
 242                                 if (fgets(linebuf, sizeof linebuf, stdin)
 243                                     == NULL) {
 244                                         if (ferror(stdin))
 245                                                 err(3, "x.c");
 246                                         return(-1);
 247                                 }
 248                                 cp = linebuf;
 249                                 continue;
 250                         }
 251                         for (tp = tmp; (ch = *tp++); tp++)
 252                                 if (c == ch) {
 253                                         c = *tp;
 254                                         goto gotc;
 255                                 }
 256                         if (!octdigit(c)) {
 257                                 *dp++ = '\\';
 258                                 break;
 259                         }
 260                         c -= '0';
 261                         if (!octdigit(*cp))
 262                                 break;
 263                         c <<= 3, c += *cp++ - '0';
 264                         if (!octdigit(*cp))
 265                                 break;
 266                         c <<= 3, c += *cp++ - '0';
 267                         break;
 268                 }
 269 gotc:
 270                 *dp++ = c;
 271         }
 272 out:
 273         *cpp = --cp;
 274         *dp = 0;
 275         return (hashit(dbuf, 1));
 276 }
 277
 278 int
 279 octdigit(char c)
 280 {
 281         return (isdigit(c) && c != '8' && c != '9');
 282 }
 283
 284 void
 285 inithash(void)
 286 {
 287         char buf[BUFSIZ];
 288         FILE *mesgread = fopen(strings, "r");
 289
 290         if (mesgread == NULL)
 291                 return;
 292         for (;;) {
 293                 mesgpt = tellpt;
 294                 if (fgetNUL(buf, sizeof buf, mesgread) == 0)
 295                         break;
 296                 ignore(hashit(buf, 0));
 297         }
 298         ignore(fclose(mesgread));
 299 }
 300
 301 int
 302 fgetNUL(char *obuf, int rmdr, FILE *file)
 303 {
 304         int c;
 305         char *buf = obuf;
 306
 307         while (--rmdr > 0 && (c = xgetc(file)) != 0 && c != EOF)
 308                 *buf++ = c;
 309         *buf++ = 0;
 310         return ((feof(file) || ferror(file)) ? 0 : 1);
 311 }
 312
 313 int
 314 xgetc(FILE *file)
 315 {
 316
 317         tellpt++;
 318         return (getc(file));
 319 }
 320
 321 #define BUCKETS 128
 322
 323 struct  hash {
 324         off_t   hpt;
 325         char    *hstr;
 326         struct  hash *hnext;
 327         short   hnew;
 328 } bucket[BUCKETS];
 329
 330 off_t
 331 hashit(char *str, int new)
 332 {
 333         int i;
 334         struct hash *hp, *hp0;
 335
 336         hp = hp0 = &bucket[lastchr(str) & 0177];
 337         while (hp->hnext) {
 338                 hp = hp->hnext;
 339                 i = istail(str, hp->hstr);
 340                 if (i >= 0)
 341                         return (hp->hpt + i);
 342         }
 343         if ((hp = (struct hash *) calloc(1, sizeof (*hp))) == NULL)
 344                 errx(8, "calloc");
 345         hp->hpt = mesgpt;
 346         if (!(hp->hstr = strdup(str)))
 347                 err(1, NULL);
 348         mesgpt += strlen(hp->hstr) + 1;
 349         hp->hnext = hp0->hnext;
 350         hp->hnew = new;
 351         hp0->hnext = hp;
 352         return (hp->hpt);
 353 }
 354
 355 void
 356 flushsh(void)
 357 {
 358         int i;
 359         struct hash *hp;
 360         FILE *mesgwrit;
 361         int old = 0, new = 0;
 362
 363         for (i = 0; i < BUCKETS; i++)
 364                 for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext)
 365                         if (hp->hnew)
 366                                 new++;
 367                         else
 368                                 old++;
 369         if (new == 0 && old != 0)
 370                 return;
 371         mesgwrit = fopen(strings, old ? "r+" : "w");
 372         if (mesgwrit == NULL)
 373                 err(4, "%s", strings);
 374         for (i = 0; i < BUCKETS; i++)
 375                 for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext) {
 376                         found(hp->hnew, hp->hpt, hp->hstr);
 377                         if (hp->hnew) {
 378                                 fseek(mesgwrit, hp->hpt, 0);
 379                                 ignore(fwrite(hp->hstr, strlen(hp->hstr) + 1, 1, mesgwrit));
 380                                 if (ferror(mesgwrit))
 381                                         err(4, "%s", strings);
 382                         }
 383                 }
 384         if (fclose(mesgwrit) == EOF)
 385                 err(4, "%s", strings);
 386 }
 387
 388 void
 389 found(int new, off_t off, char *str)
 390 {
 391         if (vflg == 0)
 392                 return;
 393         if (!new)
 394                 fprintf(stderr, "found at %d:", (int) off);
 395         else
 396                 fprintf(stderr, "new at %d:", (int) off);
 397         prstr(str);
 398         fprintf(stderr, "\n");
 399 }
 400
 401 void
 402 prstr(char *cp)
 403 {
 404         int c;
 405
 406         while ((c = (*cp++ & 0377)))
 407                 if (c < ' ')
 408                         fprintf(stderr, "^%c", c + '`');
 409                 else if (c == 0177)
 410                         fprintf(stderr, "^?");
 411                 else if (c > 0200)
 412                         fprintf(stderr, "\\%03o", c);
 413                 else
 414                         fprintf(stderr, "%c", c);
 415 }
 416
 417 void
 418 xsdotc(void)
 419 {
 420         FILE *strf = fopen(strings, "r");
 421         FILE *xdotcf;
 422
 423         if (strf == NULL)
 424                 err(5, "%s", strings);
 425         xdotcf = fopen("xs.c", "w");
 426         if (xdotcf == NULL)
 427                 err(6, "xs.c");
 428         fprintf(xdotcf, "char\txstr[] = {\n");
 429         for (;;) {
 430                 int i, c;
 431
 432                 for (i = 0; i < 8; i++) {
 433                         c = getc(strf);
 434                         if (ferror(strf)) {
 435                                 warn("%s", strings);
 436                                 onintr(0);
 437                         }
 438                         if (feof(strf)) {
 439                                 fprintf(xdotcf, "\n");
 440                                 goto out;
 441                         }
 442                         fprintf(xdotcf, "0x%02x,", c);
 443                 }
 444                 fprintf(xdotcf, "\n");
 445         }
 446 out:
 447         fprintf(xdotcf, "};\n");
 448         ignore(fclose(xdotcf));
 449         ignore(fclose(strf));
 450 }
 451
 452 char
 453 lastchr(char *cp)
 454 {
 455
 456         while (cp[0] && cp[1])
 457                 cp++;
 458         return (*cp);
 459 }
 460
 461 int
 462 istail(char *str, char *of)
 463 {
 464         int d = strlen(of) - strlen(str);
 465
 466         if (d < 0 || strcmp(&of[d], str) != 0)
 467                 return (-1);
 468         return (d);
 469 }
 470
 471 void
 472 onintr(int dummy __unused)
 473 {
 474
 475         ignore(signal(SIGINT, SIG_IGN));
 476         if (strings[0] == '/')
 477                 ignore(unlink(strings));
 478         ignore(unlink("x.c"));
 479         ignore(unlink("xs.c"));
 480         exit(7);
 481 }