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 1.4.2.1 2002/11/16 01:01:21 tjr Exp $
  36  * $DragonFly: src/usr.bin/xstr/xstr.c,v 1.3 2005/07/26 20:17:31 liamfoy Exp $
  37  */
  38
  39 #include <sys/types.h>
  40
  41 #include <ctype.h>
  42 #include <err.h>
  43 #include <stdio.h>
  44 #include <stdlib.h>
  45 #include <signal.h>
  46 #include <string.h>
  47 #include <unistd.h>
  48
  49 #include "pathnames.h"
  50
  51 /*
  52  * xstr - extract and hash strings in a C program
  53  *
  54  * Bill Joy UCB
  55  * November, 1978
  56  */
  57
  58 #define ignore(a)       ((void) a)
  59
  60 off_t   tellpt;
  61
  62 off_t   mesgpt;
  63 char    cstrings[] =    "strings";
  64 char    *strings =      cstrings;
  65
  66 int     cflg;
  67 int     vflg;
  68 int     readstd;
  69
  70 char lastchr(char *);
  71
  72 int fgetNUL(char *, int, FILE *);
  73 int istail(char *, char *);
  74 int octdigit(char);
  75 int xgetc(FILE *);
  76
  77 off_t hashit(char *, int);
  78 off_t yankstr(char **);
  79
  80 static void usage(void);
  81
  82 void flushsh(void);
  83 void found(int, off_t, char *);
  84 void inithash(void);
  85 void onintr(int);
  86 void process(const char *);
  87 void prstr(char *);
  88 void xsdotc(void);
  89
  90 int
  91 main(int argc, char *argv[])
  92 {
  93         int c;
  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 = mktemp(strdup(_PATH_TMP));
 118         while (readstd || argc > 0) {
 119                 if (freopen("x.c", "w", stdout) == NULL)
 120                         err(1, "x.c");
 121                 if (!readstd && freopen(argv[0], "r", stdin) == NULL)
 122                         err(2, "%s", argv[0]);
 123                 process("x.c");
 124                 if (readstd == 0)
 125                         argc--, argv++;
 126                 else
 127                         readstd = 0;
 128         };
 129         flushsh();
 130         if (cflg == 0)
 131                 xsdotc();
 132         if (strings[0] == '/')
 133                 ignore(unlink(strings));
 134         exit(0);
 135 }
 136
 137 static void
 138 usage(void)
 139 {
 140         fprintf(stderr, "usage: xstr [-v] [-c] [-] [name ...]\n");
 141         exit (1);
 142 }
 143
 144 char linebuf[BUFSIZ];
 145
 146 void
 147 process(const char *name)
 148 {
 149         char *cp;
 150         int c;
 151         int incomm = 0;
 152         int ret;
 153
 154         printf("extern char\txstr[];\n");
 155         for (;;) {
 156                 if (fgets(linebuf, sizeof linebuf, stdin) == NULL) {
 157                         if (ferror(stdin))
 158                                 err(3, "%s", name);
 159                         break;
 160                 }
 161                 if (linebuf[0] == '#') {
 162                         if (linebuf[1] == ' ' && isdigit(linebuf[2]))
 163                                 printf("#line%s", &linebuf[1]);
 164                         else
 165                                 printf("%s", linebuf);
 166                         continue;
 167                 }
 168                 for (cp = linebuf; (c = *cp++);) switch (c) {
 169
 170                 case '"':
 171                         if (incomm)
 172                                 goto def;
 173                         if ((ret = (int) yankstr(&cp)) == -1)
 174                                 goto out;
 175                         printf("(&xstr[%d])", ret);
 176                         break;
 177
 178                 case '\'':
 179                         if (incomm)
 180                                 goto def;
 181                         putchar(c);
 182                         if (*cp)
 183                                 putchar(*cp++);
 184                         break;
 185
 186                 case '/':
 187                         if (incomm || *cp != '*')
 188                                 goto def;
 189                         incomm = 1;
 190                         cp++;
 191                         printf("/*");
 192                         continue;
 193
 194                 case '*':
 195                         if (incomm && *cp == '/') {
 196                                 incomm = 0;
 197                                 cp++;
 198                                 printf("*/");
 199                                 continue;
 200                         }
 201                         goto def;
 202
 203 def:
 204                 default:
 205                         putchar(c);
 206                         break;
 207                 }
 208         }
 209 out:
 210         if (ferror(stdout))
 211                 warn("x.c"), onintr(0);
 212 }
 213
 214 off_t
 215 yankstr(char **cpp)
 216 {
 217         char *cp = *cpp;
 218         int c, ch;
 219         char dbuf[BUFSIZ];
 220         char *dp = dbuf;
 221         char *tp;
 222         static char tmp[] = "b\bt\tr\rn\nf\f\\\\\"\"";
 223
 224         while ((c = *cp++)) {
 225                 if (dp == dbuf + sizeof(dbuf) - 3)
 226                         errx(1, "message too long");
 227                 switch (c) {
 228
 229                 case '"':
 230                         cp++;
 231                         goto out;
 232
 233                 case '\\':
 234                         c = *cp++;
 235                         if (c == 0)
 236                                 break;
 237                         if (c == '\n') {
 238                                 if (fgets(linebuf, sizeof linebuf, stdin)
 239                                     == NULL) {
 240                                         if (ferror(stdin))
 241                                                 err(3, "x.c");
 242                                         return(-1);
 243                                 }
 244                                 cp = linebuf;
 245                                 continue;
 246                         }
 247                         for (tp = tmp; (ch = *tp++); tp++)
 248                                 if (c == ch) {
 249                                         c = *tp;
 250                                         goto gotc;
 251                                 }
 252                         if (!octdigit(c)) {
 253                                 *dp++ = '\\';
 254                                 break;
 255                         }
 256                         c -= '0';
 257                         if (!octdigit(*cp))
 258                                 break;
 259                         c <<= 3, c += *cp++ - '0';
 260                         if (!octdigit(*cp))
 261                                 break;
 262                         c <<= 3, c += *cp++ - '0';
 263                         break;
 264                 }
 265 gotc:
 266                 *dp++ = c;
 267         }
 268 out:
 269         *cpp = --cp;
 270         *dp = 0;
 271         return (hashit(dbuf, 1));
 272 }
 273
 274 int
 275 octdigit(char c)
 276 {
 277         return (isdigit(c) && c != '8' && c != '9');
 278 }
 279
 280 void
 281 inithash(void)
 282 {
 283         char buf[BUFSIZ];
 284         FILE *mesgread = fopen(strings, "r");
 285
 286         if (mesgread == NULL)
 287                 return;
 288         for (;;) {
 289                 mesgpt = tellpt;
 290                 if (fgetNUL(buf, sizeof buf, mesgread) == 0)
 291                         break;
 292                 ignore(hashit(buf, 0));
 293         }
 294         ignore(fclose(mesgread));
 295 }
 296
 297 int
 298 fgetNUL(char *obuf, int rmdr, FILE *file)
 299 {
 300         int c;
 301         char *buf = obuf;
 302
 303         while (--rmdr > 0 && (c = xgetc(file)) != 0 && c != EOF)
 304                 *buf++ = c;
 305         *buf++ = 0;
 306         return ((feof(file) || ferror(file)) ? 0 : 1);
 307 }
 308
 309 int
 310 xgetc(FILE *file)
 311 {
 312
 313         tellpt++;
 314         return (getc(file));
 315 }
 316
 317 #define BUCKETS 128
 318
 319 struct  hash {
 320         off_t   hpt;
 321         char    *hstr;
 322         struct  hash *hnext;
 323         short   hnew;
 324 } bucket[BUCKETS];
 325
 326 off_t
 327 hashit(char *str, int new)
 328 {
 329         int i;
 330         struct hash *hp, *hp0;
 331
 332         hp = hp0 = &bucket[lastchr(str) & 0177];
 333         while (hp->hnext) {
 334                 hp = hp->hnext;
 335                 i = istail(str, hp->hstr);
 336                 if (i >= 0)
 337                         return (hp->hpt + i);
 338         }
 339         if ((hp = (struct hash *) calloc(1, sizeof (*hp))) == NULL)
 340                 errx(8, "calloc");
 341         hp->hpt = mesgpt;
 342         if (!(hp->hstr = strdup(str)))
 343                 err(1, NULL);
 344         mesgpt += strlen(hp->hstr) + 1;
 345         hp->hnext = hp0->hnext;
 346         hp->hnew = new;
 347         hp0->hnext = hp;
 348         return (hp->hpt);
 349 }
 350
 351 void
 352 flushsh(void)
 353 {
 354         int i;
 355         struct hash *hp;
 356         FILE *mesgwrit;
 357         int old = 0, new = 0;
 358
 359         for (i = 0; i < BUCKETS; i++)
 360                 for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext)
 361                         if (hp->hnew)
 362                                 new++;
 363                         else
 364                                 old++;
 365         if (new == 0 && old != 0)
 366                 return;
 367         mesgwrit = fopen(strings, old ? "r+" : "w");
 368         if (mesgwrit == NULL)
 369                 err(4, "%s", strings);
 370         for (i = 0; i < BUCKETS; i++)
 371                 for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext) {
 372                         found(hp->hnew, hp->hpt, hp->hstr);
 373                         if (hp->hnew) {
 374                                 fseek(mesgwrit, hp->hpt, 0);
 375                                 ignore(fwrite(hp->hstr, strlen(hp->hstr) + 1, 1, mesgwrit));
 376                                 if (ferror(mesgwrit))
 377                                         err(4, "%s", strings);
 378                         }
 379                 }
 380         if (fclose(mesgwrit) == EOF)
 381                 err(4, "%s", strings);
 382 }
 383
 384 void
 385 found(int new, off_t off, char *str)
 386 {
 387         if (vflg == 0)
 388                 return;
 389         if (!new)
 390                 fprintf(stderr, "found at %d:", (int) off);
 391         else
 392                 fprintf(stderr, "new at %d:", (int) off);
 393         prstr(str);
 394         fprintf(stderr, "\n");
 395 }
 396
 397 void
 398 prstr(char *cp)
 399 {
 400         int c;
 401
 402         while ((c = (*cp++ & 0377)))
 403                 if (c < ' ')
 404                         fprintf(stderr, "^%c", c + '`');
 405                 else if (c == 0177)
 406                         fprintf(stderr, "^?");
 407                 else if (c > 0200)
 408                         fprintf(stderr, "\\%03o", c);
 409                 else
 410                         fprintf(stderr, "%c", c);
 411 }
 412
 413 void
 414 xsdotc(void)
 415 {
 416         FILE *strf = fopen(strings, "r");
 417         FILE *xdotcf;
 418
 419         if (strf == NULL)
 420                 err(5, "%s", strings);
 421         xdotcf = fopen("xs.c", "w");
 422         if (xdotcf == NULL)
 423                 err(6, "xs.c");
 424         fprintf(xdotcf, "char\txstr[] = {\n");
 425         for (;;) {
 426                 int i, c;
 427
 428                 for (i = 0; i < 8; i++) {
 429                         c = getc(strf);
 430                         if (ferror(strf)) {
 431                                 warn("%s", strings);
 432                                 onintr(0);
 433                         }
 434                         if (feof(strf)) {
 435                                 fprintf(xdotcf, "\n");
 436                                 goto out;
 437                         }
 438                         fprintf(xdotcf, "0x%02x,", c);
 439                 }
 440                 fprintf(xdotcf, "\n");
 441         }
 442 out:
 443         fprintf(xdotcf, "};\n");
 444         ignore(fclose(xdotcf));
 445         ignore(fclose(strf));
 446 }
 447
 448 char
 449 lastchr(char *cp)
 450 {
 451
 452         while (cp[0] && cp[1])
 453                 cp++;
 454         return (*cp);
 455 }
 456
 457 int
 458 istail(char *str, char *of)
 459 {
 460         int d = strlen(of) - strlen(str);
 461
 462         if (d < 0 || strcmp(&of[d], str) != 0)
 463                 return (-1);
 464         return (d);
 465 }
 466
 467 void
 468 onintr(int dummy __unused)
 469 {
 470
 471         ignore(signal(SIGINT, SIG_IGN));
 472         if (strings[0] == '/')
 473                 ignore(unlink(strings));
 474         ignore(unlink("x.c"));
 475         ignore(unlink("xs.c"));
 476         exit(7);
 477 }