| 1 | /* Copyright (c) 1979 Regents of the University of California */ |
| 2 | #include <stdio.h> |
| 3 | #include <ctype.h> |
| 4 | #include <sys/types.h> |
| 5 | #include <signal.h> |
| 6 | |
| 7 | int lseek(); /* Chicanery */ |
| 8 | /* |
| 9 | * xstr - extract and hash strings in a C program |
| 10 | * |
| 11 | * Bill Joy UCB |
| 12 | * November, 1978 |
| 13 | */ |
| 14 | |
| 15 | #define ignore(a) Ignore((char *) a) |
| 16 | |
| 17 | char *calloc(); |
| 18 | off_t tellpt; |
| 19 | off_t hashit(); |
| 20 | char *mktemp(); |
| 21 | int onintr(); |
| 22 | char *savestr(); |
| 23 | char *strcat(); |
| 24 | char *strcpy(); |
| 25 | off_t yankstr(); |
| 26 | |
| 27 | off_t mesgpt; |
| 28 | char *strings = "strings"; |
| 29 | |
| 30 | int cflg; |
| 31 | int vflg; |
| 32 | int readstd; |
| 33 | |
| 34 | main(argc, argv) |
| 35 | int argc; |
| 36 | char *argv[]; |
| 37 | { |
| 38 | |
| 39 | argc--, argv++; |
| 40 | while (argc > 0 && argv[0][0] == '-') { |
| 41 | register char *cp = &(*argv++)[1]; |
| 42 | |
| 43 | argc--; |
| 44 | if (*cp == 0) { |
| 45 | readstd++; |
| 46 | continue; |
| 47 | } |
| 48 | do switch (*cp++) { |
| 49 | |
| 50 | case 'c': |
| 51 | cflg++; |
| 52 | continue; |
| 53 | |
| 54 | case 'v': |
| 55 | vflg++; |
| 56 | continue; |
| 57 | |
| 58 | default: |
| 59 | fprintf(stderr, "usage: xstr [ -v ] [ -c ] [ - ] [ name ... ]\n"); |
| 60 | } while (*cp); |
| 61 | } |
| 62 | if (signal(SIGINT, SIG_IGN) == SIG_DFL) |
| 63 | signal(SIGINT, onintr); |
| 64 | if (cflg || argc == 0 && !readstd) |
| 65 | inithash(); |
| 66 | else |
| 67 | strings = mktemp(savestr("/tmp/xstrXXXXXX")); |
| 68 | while (readstd || argc > 0) { |
| 69 | if (freopen("x.c", "w", stdout) == NULL) |
| 70 | perror("x.c"), exit(1); |
| 71 | if (!readstd && freopen(argv[0], "r", stdin) == NULL) |
| 72 | perror(argv[0]), exit(2); |
| 73 | process("x.c"); |
| 74 | if (readstd == 0) |
| 75 | argc--, argv++; |
| 76 | else |
| 77 | readstd = 0; |
| 78 | }; |
| 79 | flushsh(); |
| 80 | if (cflg == 0) |
| 81 | xsdotc(); |
| 82 | if (strings[0] == '/') |
| 83 | ignore(unlink(strings)); |
| 84 | exit(0); |
| 85 | } |
| 86 | |
| 87 | process(name) |
| 88 | char *name; |
| 89 | { |
| 90 | char *cp; |
| 91 | char linebuf[BUFSIZ]; |
| 92 | register int c; |
| 93 | register int incomm = 0; |
| 94 | |
| 95 | printf("char\txstr[];\n"); |
| 96 | for (;;) { |
| 97 | if (fgets(linebuf, sizeof linebuf, stdin) == NULL) { |
| 98 | if (ferror(stdin)) { |
| 99 | perror(name); |
| 100 | exit(3); |
| 101 | } |
| 102 | break; |
| 103 | } |
| 104 | if (linebuf[0] == '#') { |
| 105 | if (linebuf[1] == ' ' && isdigit(linebuf[2])) |
| 106 | printf("#line%s", &linebuf[1]); |
| 107 | else |
| 108 | printf("%s", linebuf); |
| 109 | continue; |
| 110 | } |
| 111 | for (cp = linebuf; c = *cp++;) switch (c) { |
| 112 | |
| 113 | case '"': |
| 114 | if (incomm) |
| 115 | goto def; |
| 116 | printf("(&xstr[%d])", (int) yankstr(&cp)); |
| 117 | break; |
| 118 | |
| 119 | case '\'': |
| 120 | if (incomm) |
| 121 | goto def; |
| 122 | putchar(c); |
| 123 | if (*cp) |
| 124 | putchar(*cp++); |
| 125 | break; |
| 126 | |
| 127 | case '/': |
| 128 | if (incomm || *cp != '*') |
| 129 | goto def; |
| 130 | incomm = 1; |
| 131 | cp++; |
| 132 | printf("/*"); |
| 133 | continue; |
| 134 | |
| 135 | case '*': |
| 136 | if (incomm && *cp == '/') { |
| 137 | incomm = 0; |
| 138 | cp++; |
| 139 | printf("*/"); |
| 140 | continue; |
| 141 | } |
| 142 | goto def; |
| 143 | |
| 144 | def: |
| 145 | default: |
| 146 | putchar(c); |
| 147 | break; |
| 148 | } |
| 149 | } |
| 150 | if (ferror(stdout)) |
| 151 | perror("x.c"), onintr(); |
| 152 | } |
| 153 | |
| 154 | off_t |
| 155 | yankstr(cpp) |
| 156 | register char **cpp; |
| 157 | { |
| 158 | register char *cp = *cpp; |
| 159 | register int c, ch; |
| 160 | char dbuf[BUFSIZ]; |
| 161 | register char *dp = dbuf; |
| 162 | register char *tp; |
| 163 | |
| 164 | while (c = *cp++) { |
| 165 | switch (c) { |
| 166 | |
| 167 | case '"': |
| 168 | cp++; |
| 169 | goto out; |
| 170 | |
| 171 | case '\\': |
| 172 | c = *cp++; |
| 173 | if (c == 0) |
| 174 | break; |
| 175 | if (c == '\n') |
| 176 | continue; |
| 177 | for (tp = "b\bt\tr\rn\nf\f\\\\\"\""; ch = *tp++; tp++) |
| 178 | if (c == ch) { |
| 179 | c = *tp; |
| 180 | goto gotc; |
| 181 | } |
| 182 | if (!octdigit(c)) { |
| 183 | *dp++ = '\\'; |
| 184 | break; |
| 185 | } |
| 186 | c -= '0'; |
| 187 | if (!octdigit(*cp)) |
| 188 | break; |
| 189 | c <<= 3, c += *cp++ - '0'; |
| 190 | if (!octdigit(*cp)) |
| 191 | break; |
| 192 | c <<= 3, c += *cp++ - '0'; |
| 193 | break; |
| 194 | } |
| 195 | gotc: |
| 196 | *dp++ = c; |
| 197 | } |
| 198 | out: |
| 199 | *cpp = --cp; |
| 200 | *dp = 0; |
| 201 | return (hashit(dbuf, 1)); |
| 202 | } |
| 203 | |
| 204 | octdigit(c) |
| 205 | char c; |
| 206 | { |
| 207 | |
| 208 | return (isdigit(c) && c != '8' && c != '9'); |
| 209 | } |
| 210 | |
| 211 | inithash() |
| 212 | { |
| 213 | char buf[BUFSIZ]; |
| 214 | register FILE *mesgread = fopen(strings, "r"); |
| 215 | |
| 216 | if (mesgread == NULL) |
| 217 | return; |
| 218 | for (;;) { |
| 219 | mesgpt = tellpt; |
| 220 | if (fgetNUL(buf, sizeof buf, mesgread) == NULL) |
| 221 | break; |
| 222 | ignore(hashit(buf, 0)); |
| 223 | } |
| 224 | ignore(fclose(mesgread)); |
| 225 | } |
| 226 | |
| 227 | fgetNUL(obuf, rmdr, file) |
| 228 | char *obuf; |
| 229 | register int rmdr; |
| 230 | FILE *file; |
| 231 | { |
| 232 | register c; |
| 233 | register char *buf = obuf; |
| 234 | |
| 235 | while (--rmdr > 0 && (c = xgetc(file)) != 0 && c != EOF) |
| 236 | *buf++ = c; |
| 237 | *buf++ = 0; |
| 238 | return ((feof(file) || ferror(file)) ? NULL : 1); |
| 239 | } |
| 240 | |
| 241 | xgetc(file) |
| 242 | FILE *file; |
| 243 | { |
| 244 | |
| 245 | tellpt++; |
| 246 | return (getc(file)); |
| 247 | } |
| 248 | |
| 249 | #define BUCKETS 128 |
| 250 | |
| 251 | struct hash { |
| 252 | off_t hpt; |
| 253 | char *hstr; |
| 254 | struct hash *hnext; |
| 255 | short hnew; |
| 256 | } bucket[BUCKETS]; |
| 257 | |
| 258 | off_t |
| 259 | hashit(str, new) |
| 260 | char *str; |
| 261 | int new; |
| 262 | { |
| 263 | int i; |
| 264 | register struct hash *hp, *hp0; |
| 265 | |
| 266 | hp = hp0 = &bucket[lastchr(str) & 0177]; |
| 267 | while (hp->hnext) { |
| 268 | hp = hp->hnext; |
| 269 | i = istail(str, hp->hstr); |
| 270 | if (i >= 0) |
| 271 | return (hp->hpt + i); |
| 272 | } |
| 273 | hp = (struct hash *) calloc(1, sizeof (*hp)); |
| 274 | hp->hpt = mesgpt; |
| 275 | hp->hstr = savestr(str); |
| 276 | mesgpt += strlen(hp->hstr) + 1; |
| 277 | hp->hnext = hp0->hnext; |
| 278 | hp->hnew = new; |
| 279 | hp0->hnext = hp; |
| 280 | return (hp->hpt); |
| 281 | } |
| 282 | |
| 283 | flushsh() |
| 284 | { |
| 285 | register int i; |
| 286 | register struct hash *hp; |
| 287 | register FILE *mesgwrit; |
| 288 | register int old = 0, new = 0; |
| 289 | |
| 290 | for (i = 0; i < BUCKETS; i++) |
| 291 | for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext) |
| 292 | if (hp->hnew) |
| 293 | new++; |
| 294 | else |
| 295 | old++; |
| 296 | if (new == 0 && old != 0) |
| 297 | return; |
| 298 | mesgwrit = fopen(strings, old ? "a" : "w"); |
| 299 | for (i = 0; i < BUCKETS; i++) |
| 300 | for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext) { |
| 301 | found(hp->hnew, hp->hpt, hp->hstr); |
| 302 | if (hp->hnew) { |
| 303 | fseek(mesgwrit, hp->hpt, 0); |
| 304 | ignore(fwrite(hp->hstr, strlen(hp->hstr) + 1, 1, mesgwrit)); |
| 305 | if (ferror(mesgwrit)) |
| 306 | perror(strings), exit(4); |
| 307 | } |
| 308 | } |
| 309 | ignore(fclose(mesgwrit)); |
| 310 | } |
| 311 | |
| 312 | found(new, off, str) |
| 313 | int new; |
| 314 | off_t off; |
| 315 | char *str; |
| 316 | { |
| 317 | register char *cp; |
| 318 | |
| 319 | if (vflg == 0) |
| 320 | return; |
| 321 | if (!new) |
| 322 | fprintf(stderr, "found at %d:", (int) off); |
| 323 | else |
| 324 | fprintf(stderr, "new at %d:", (int) off); |
| 325 | prstr(str); |
| 326 | fprintf(stderr, "\n"); |
| 327 | } |
| 328 | |
| 329 | prstr(cp) |
| 330 | register char *cp; |
| 331 | { |
| 332 | register int c; |
| 333 | |
| 334 | while (c = (*cp++ & 0377)) |
| 335 | if (c < ' ') |
| 336 | fprintf(stderr, "^%c", c + '`'); |
| 337 | else if (c == 0177) |
| 338 | fprintf(stderr, "^?"); |
| 339 | else if (c > 0200) |
| 340 | fprintf(stderr, "\\%03o", c); |
| 341 | else |
| 342 | fprintf(stderr, "%c", c); |
| 343 | } |
| 344 | |
| 345 | xsdotc() |
| 346 | { |
| 347 | register FILE *strf = fopen(strings, "r"); |
| 348 | register FILE *xdotcf; |
| 349 | |
| 350 | if (strf == NULL) |
| 351 | perror(strings), exit(5); |
| 352 | xdotcf = fopen("xs.c", "w"); |
| 353 | if (xdotcf == NULL) |
| 354 | perror("xs.c"), exit(6); |
| 355 | fprintf(xdotcf, "char\txstr[] = {\n"); |
| 356 | for (;;) { |
| 357 | register int i, c; |
| 358 | |
| 359 | for (i = 0; i < 8; i++) { |
| 360 | c = getc(strf); |
| 361 | if (ferror(strf)) { |
| 362 | perror(strings); |
| 363 | onintr(); |
| 364 | } |
| 365 | if (feof(strf)) { |
| 366 | fprintf(xdotcf, "\n"); |
| 367 | goto out; |
| 368 | } |
| 369 | fprintf(xdotcf, "0x%02x,", c); |
| 370 | } |
| 371 | fprintf(xdotcf, "\n"); |
| 372 | } |
| 373 | out: |
| 374 | fprintf(xdotcf, "};\n"); |
| 375 | ignore(fclose(xdotcf)); |
| 376 | ignore(fclose(strf)); |
| 377 | } |
| 378 | |
| 379 | char * |
| 380 | savestr(cp) |
| 381 | register char *cp; |
| 382 | { |
| 383 | register char *dp = (char *) calloc(1, strlen(cp) + 1); |
| 384 | |
| 385 | return (strcpy(dp, cp)); |
| 386 | } |
| 387 | |
| 388 | Ignore(a) |
| 389 | char *a; |
| 390 | { |
| 391 | |
| 392 | a = a; |
| 393 | } |
| 394 | |
| 395 | ignorf(a) |
| 396 | int (*a)(); |
| 397 | { |
| 398 | |
| 399 | a = a; |
| 400 | } |
| 401 | |
| 402 | lastchr(cp) |
| 403 | register char *cp; |
| 404 | { |
| 405 | |
| 406 | while (cp[0] && cp[1]) |
| 407 | cp++; |
| 408 | return (*cp); |
| 409 | } |
| 410 | |
| 411 | istail(str, of) |
| 412 | register char *str, *of; |
| 413 | { |
| 414 | register int d = strlen(of) - strlen(str); |
| 415 | |
| 416 | if (d < 0 || strcmp(&of[d], str) != 0) |
| 417 | return (-1); |
| 418 | return (d); |
| 419 | } |
| 420 | |
| 421 | onintr() |
| 422 | { |
| 423 | |
| 424 | ignorf(signal(SIGINT, SIG_IGN)); |
| 425 | if (strings[0] == '/') |
| 426 | ignore(unlink(strings)); |
| 427 | ignore(unlink("x.c")); |
| 428 | ignore(unlink("xs.c")); |
| 429 | exit(7); |
| 430 | } |