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  * @(#) Copyright (c) 1980, 1993 The Regents of the University of California.  All rights reserved.
  34  * @(#)xstr.c   8.1 (Berkeley) 6/9/93
  35  * $FreeBSD: src/usr.bin/xstr/xstr.c,v 1.11 2008/05/13 09:42:03 kevlo Exp $
  36  */
  37
  38 #include <sys/types.h>
  39
  40 #include <ctype.h>
  41 #include <err.h>
  42 #include <stdio.h>
  43 #include <stdlib.h>
  44 #include <signal.h>
  45 #include <string.h>
  46 #include <unistd.h>
  47
  48 #include "pathnames.h"
  49
  50 /*
  51  * xstr - extract and hash strings in a C program
  52  *
  53  * Bill Joy UCB
  54  * November, 1978
  55  */
  56
  57 #define ignore(a)       ((void) a)
  58
  59 off_t   tellpt;
  60
  61 off_t   mesgpt;
  62 char    cstrings[] =    "strings";
  63 char    *strings =      cstrings;
  64
  65 int     cflg;
  66 int     vflg;
  67 int     readstd;
  68
  69 char lastchr(char *);
  70
  71 int fgetNUL(char *, int, FILE *);
  72 int istail(char *, char *);
  73 int octdigit(char);
  74 int xgetc(FILE *);
  75
  76 off_t hashit(char *, int);
  77 off_t yankstr(char **);
  78
  79 static void usage(void);
  80
  81 void flushsh(void);
  82 void found(int, off_t, char *);
  83 void inithash(void);
  84 void onintr(int);
  85 void process(const char *);
  86 void prstr(char *);
  87 void xsdotc(void);
  88
  89 int
  90 main(int argc, char *argv[])
  91 {
  92         int c;
  93         int fdesc;
  94
  95         while ((c = getopt(argc, argv, "-cv")) != -1)
  96                 switch (c) {
  97                 case '-':
  98                         readstd++;
  99                         break;
 100                 case 'c':
 101                         cflg++;
 102                         break;
 103                 case 'v':
 104                         vflg++;
 105                         break;
 106                 default:
 107                         usage();
 108                 }
 109         argc -= optind;
 110         argv += optind;
 111
 112         if (signal(SIGINT, SIG_IGN) == SIG_DFL)
 113                 signal(SIGINT, onintr);
 114         if (cflg || (argc == 0 && !readstd))
 115                 inithash();
 116         else {
 117                 strings = strdup(_PATH_TMP);
 118                 if (strings == NULL)
 119                         err(1, "strdup() failed");
 120                 fdesc = mkstemp(strings);
 121                 if (fdesc == -1)
 122                         err(1, "Unable to create temporary file");
 123                 close(fdesc);
 124         }
 125
 126         while (readstd || argc > 0) {
 127                 if (freopen("x.c", "w", stdout) == NULL)
 128                         err(1, "x.c");
 129                 if (!readstd && freopen(argv[0], "r", stdin) == NULL)
 130                         err(2, "%s", argv[0]);
 131                 process("x.c");
 132                 if (readstd == 0)
 133                         argc--, argv++;
 134                 else
 135                         readstd = 0;
 136         };
 137         flushsh();
 138         if (cflg == 0)
 139                 xsdotc();
 140         if (strings[0] == '/')
 141                 ignore(unlink(strings));
 142         exit(0);
 143 }
 144
 145 static void
 146 usage(void)
 147 {
 148         fprintf(stderr, "usage: xstr [-cv] [-] [file ...]\n");
 149         exit (1);
 150 }
 151
 152 char linebuf[BUFSIZ];
 153
 154 void
 155 process(const char *name)
 156 {
 157         char *cp;
 158         int c;
 159         int incomm = 0;
 160         int ret;
 161
 162         printf("extern char\txstr[];\n");
 163         for (;;) {
 164                 if (fgets(linebuf, sizeof linebuf, stdin) == NULL) {
 165                         if (ferror(stdin))
 166                                 err(3, "%s", name);
 167                         break;
 168                 }
 169                 if (linebuf[0] == '#') {
 170                         if (linebuf[1] == ' ' && isdigit(linebuf[2]))
 171                                 printf("#line%s", &linebuf[1]);
 172                         else
 173                                 printf("%s", linebuf);
 174                         continue;
 175                 }
 176                 for (cp = linebuf; (c = *cp++);) switch (c) {
 177
 178                 case '"':
 179                         if (incomm)
 180                                 goto def;
 181                         if ((ret = (int) yankstr(&cp)) == -1)
 182                                 goto out;
 183                         printf("(&xstr[%d])", ret);
 184                         break;
 185
 186                 case '\'':
 187                         if (incomm)
 188                                 goto def;
 189                         putchar(c);
 190                         if (*cp)
 191                                 putchar(*cp++);
 192                         break;
 193
 194                 case '/':
 195                         if (incomm || *cp != '*')
 196                                 goto def;
 197                         incomm = 1;
 198                         cp++;
 199                         printf("/*");
 200                         continue;
 201
 202                 case '*':
 203                         if (incomm && *cp == '/') {
 204                                 incomm = 0;
 205                                 cp++;
 206                                 printf("*/");
 207                                 continue;
 208                         }
 209                         goto def;
 210
 211 def:
 212                 default:
 213                         putchar(c);
 214                         break;
 215                 }
 216         }
 217 out:
 218         if (ferror(stdout))
 219                 warn("x.c"), onintr(0);
 220 }
 221
 222 off_t
 223 yankstr(char **cpp)
 224 {
 225         char *cp = *cpp;
 226         int c, ch;
 227         char dbuf[BUFSIZ];
 228         char *dp = dbuf;
 229         char *tp;
 230         static char tmp[] = "b\bt\tr\rn\nf\f\\\\\"\"";
 231
 232         while ((c = *cp++)) {
 233                 if (dp == dbuf + sizeof(dbuf) - 3)
 234                         errx(1, "message too long");
 235                 switch (c) {
 236
 237                 case '"':
 238                         cp++;
 239                         goto out;
 240
 241                 case '\\':
 242                         c = *cp++;
 243                         if (c == 0)
 244                                 break;
 245                         if (c == '\n') {
 246                                 if (fgets(linebuf, sizeof linebuf, stdin)
 247                                     == NULL) {
 248                                         if (ferror(stdin))
 249                                                 err(3, "x.c");
 250                                         return(-1);
 251                                 }
 252                                 cp = linebuf;
 253                                 continue;
 254                         }
 255                         for (tp = tmp; (ch = *tp++); tp++)
 256                                 if (c == ch) {
 257                                         c = *tp;
 258                                         goto gotc;
 259                                 }
 260                         if (!octdigit(c)) {
 261                                 *dp++ = '\\';
 262                                 break;
 263                         }
 264                         c -= '0';
 265                         if (!octdigit(*cp))
 266                                 break;
 267                         c <<= 3, c += *cp++ - '0';
 268                         if (!octdigit(*cp))
 269                                 break;
 270                         c <<= 3, c += *cp++ - '0';
 271                         break;
 272                 }
 273 gotc:
 274                 *dp++ = c;
 275         }
 276 out:
 277         *cpp = --cp;
 278         *dp = 0;
 279         return (hashit(dbuf, 1));
 280 }
 281
 282 int
 283 octdigit(char c)
 284 {
 285         return (isdigit(c) && c != '8' && c != '9');
 286 }
 287
 288 void
 289 inithash(void)
 290 {
 291         char buf[BUFSIZ];
 292         FILE *mesgread = fopen(strings, "r");
 293
 294         if (mesgread == NULL)
 295                 return;
 296         for (;;) {
 297                 mesgpt = tellpt;
 298                 if (fgetNUL(buf, sizeof buf, mesgread) == 0)
 299                         break;
 300                 ignore(hashit(buf, 0));
 301         }
 302         ignore(fclose(mesgread));
 303 }
 304
 305 int
 306 fgetNUL(char *obuf, int rmdr, FILE *file)
 307 {
 308         int c;
 309         char *buf = obuf;
 310
 311         while (--rmdr > 0 && (c = xgetc(file)) != 0 && c != EOF)
 312                 *buf++ = c;
 313         *buf++ = 0;
 314         return ((feof(file) || ferror(file)) ? 0 : 1);
 315 }
 316
 317 int
 318 xgetc(FILE *file)
 319 {
 320
 321         tellpt++;
 322         return (getc(file));
 323 }
 324
 325 #define BUCKETS 128
 326
 327 struct  hash {
 328         off_t   hpt;
 329         char    *hstr;
 330         struct  hash *hnext;
 331         short   hnew;
 332 } bucket[BUCKETS];
 333
 334 off_t
 335 hashit(char *str, int new)
 336 {
 337         int i;
 338         struct hash *hp, *hp0;
 339
 340         hp = hp0 = &bucket[lastchr(str) & 0177];
 341         while (hp->hnext) {
 342                 hp = hp->hnext;
 343                 i = istail(str, hp->hstr);
 344                 if (i >= 0)
 345                         return (hp->hpt + i);
 346         }
 347         if ((hp = (struct hash *) calloc(1, sizeof (*hp))) == NULL)
 348                 errx(8, "calloc");
 349         hp->hpt = mesgpt;
 350         if (!(hp->hstr = strdup(str)))
 351                 err(1, NULL);
 352         mesgpt += strlen(hp->hstr) + 1;
 353         hp->hnext = hp0->hnext;
 354         hp->hnew = new;
 355         hp0->hnext = hp;
 356         return (hp->hpt);
 357 }
 358
 359 void
 360 flushsh(void)
 361 {
 362         int i;
 363         struct hash *hp;
 364         FILE *mesgwrit;
 365         int old = 0, new = 0;
 366
 367         for (i = 0; i < BUCKETS; i++)
 368                 for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext)
 369                         if (hp->hnew)
 370                                 new++;
 371                         else
 372                                 old++;
 373         if (new == 0 && old != 0)
 374                 return;
 375         mesgwrit = fopen(strings, old ? "r+" : "w");
 376         if (mesgwrit == NULL)
 377                 err(4, "%s", strings);
 378         for (i = 0; i < BUCKETS; i++)
 379                 for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext) {
 380                         found(hp->hnew, hp->hpt, hp->hstr);
 381                         if (hp->hnew) {
 382                                 fseek(mesgwrit, hp->hpt, 0);
 383                                 ignore(fwrite(hp->hstr, strlen(hp->hstr) + 1, 1, mesgwrit));
 384                                 if (ferror(mesgwrit))
 385                                         err(4, "%s", strings);
 386                         }
 387                 }
 388         if (fclose(mesgwrit) == EOF)
 389                 err(4, "%s", strings);
 390 }
 391
 392 void
 393 found(int new, off_t off, char *str)
 394 {
 395         if (vflg == 0)
 396                 return;
 397         if (!new)
 398                 fprintf(stderr, "found at %d:", (int) off);
 399         else
 400                 fprintf(stderr, "new at %d:", (int) off);
 401         prstr(str);
 402         fprintf(stderr, "\n");
 403 }
 404
 405 void
 406 prstr(char *cp)
 407 {
 408         int c;
 409
 410         while ((c = (*cp++ & 0377)))
 411                 if (c < ' ')
 412                         fprintf(stderr, "^%c", c + '`');
 413                 else if (c == 0177)
 414                         fprintf(stderr, "^?");
 415                 else if (c > 0200)
 416                         fprintf(stderr, "\\%03o", c);
 417                 else
 418                         fprintf(stderr, "%c", c);
 419 }
 420
 421 void
 422 xsdotc(void)
 423 {
 424         FILE *strf = fopen(strings, "r");
 425         FILE *xdotcf;
 426
 427         if (strf == NULL)
 428                 err(5, "%s", strings);
 429         xdotcf = fopen("xs.c", "w");
 430         if (xdotcf == NULL)
 431                 err(6, "xs.c");
 432         fprintf(xdotcf, "char\txstr[] = {\n");
 433         for (;;) {
 434                 int i, c;
 435
 436                 for (i = 0; i < 8; i++) {
 437                         c = getc(strf);
 438                         if (ferror(strf)) {
 439                                 warn("%s", strings);
 440                                 onintr(0);
 441                         }
 442                         if (feof(strf)) {
 443                                 fprintf(xdotcf, "\n");
 444                                 goto out;
 445                         }
 446                         fprintf(xdotcf, "0x%02x,", c);
 447                 }
 448                 fprintf(xdotcf, "\n");
 449         }
 450 out:
 451         fprintf(xdotcf, "};\n");
 452         ignore(fclose(xdotcf));
 453         ignore(fclose(strf));
 454 }
 455
 456 char
 457 lastchr(char *cp)
 458 {
 459
 460         while (cp[0] && cp[1])
 461                 cp++;
 462         return (*cp);
 463 }
 464
 465 int
 466 istail(char *str, char *of)
 467 {
 468         int d = strlen(of) - strlen(str);
 469
 470         if (d < 0 || strcmp(&of[d], str) != 0)
 471                 return (-1);
 472         return (d);
 473 }
 474
 475 void
 476 onintr(int dummy __unused)
 477 {
 478
 479         ignore(signal(SIGINT, SIG_IGN));
 480         if (strings[0] == '/')
 481                 ignore(unlink(strings));
 482         ignore(unlink("x.c"));
 483         ignore(unlink("xs.c"));
 484         exit(7);
 485 }