| 1 | #ifndef lint |
| 2 | static char sccsid[] = "@(#)file.c 4.15 (Berkeley) %G%"; |
| 3 | #endif |
| 4 | /* |
| 5 | * file - determine type of file |
| 6 | */ |
| 7 | |
| 8 | #include <sys/param.h> |
| 9 | #include <sys/stat.h> |
| 10 | #include <stdio.h> |
| 11 | #include <ctype.h> |
| 12 | #include <a.out.h> |
| 13 | |
| 14 | extern int errno; |
| 15 | extern int sys_nerr; |
| 16 | extern char *sys_errlist[]; |
| 17 | |
| 18 | int in; |
| 19 | int i = 0; |
| 20 | char buf[BUFSIZ]; |
| 21 | char *troff[] = { /* new troff intermediate lang */ |
| 22 | "x","T","res","init","font","202","V0","p1",0}; |
| 23 | char *fort[] = { |
| 24 | "function","subroutine","common","dimension","block","integer", |
| 25 | "real","data","double",0}; |
| 26 | char *asc[] = { |
| 27 | "chmk","mov","tst","clr","jmp",0}; |
| 28 | char *c[] = { |
| 29 | "int","char","float","double","struct","extern",0}; |
| 30 | char *as[] = { |
| 31 | "globl","byte","align","text","data","comm",0}; |
| 32 | char *sh[] = { |
| 33 | "fi", "elif", "esac", "done", "export", |
| 34 | "readonly", "trap", "PATH", "HOME", 0 }; |
| 35 | char *csh[] = { |
| 36 | "alias", "breaksw", "endsw", "foreach", "limit", "onintr", |
| 37 | "repeat", "setenv", "source", "path", "home", 0 }; |
| 38 | int ifile; |
| 39 | |
| 40 | main(argc, argv) |
| 41 | char **argv; |
| 42 | { |
| 43 | FILE *fl; |
| 44 | register char *p; |
| 45 | char ap[MAXPATHLEN + 1]; |
| 46 | |
| 47 | if (argc < 2) { |
| 48 | fprintf(stderr, "usage: %s file ...\n", argv[0]); |
| 49 | exit(3); |
| 50 | } |
| 51 | |
| 52 | if (argc>1 && argv[1][0]=='-' && argv[1][1]=='f') { |
| 53 | if ((fl = fopen(argv[2], "r")) == NULL) { |
| 54 | perror(argv[2]); |
| 55 | exit(2); |
| 56 | } |
| 57 | while ((p = fgets(ap, sizeof ap, fl)) != NULL) { |
| 58 | int l = strlen(p); |
| 59 | if (l>0) |
| 60 | p[l-1] = '\0'; |
| 61 | printf("%s: ", p); |
| 62 | type(p); |
| 63 | if (ifile>=0) |
| 64 | close(ifile); |
| 65 | } |
| 66 | exit(1); |
| 67 | } |
| 68 | while(argc > 1) { |
| 69 | printf("%s: ", argv[1]); |
| 70 | type(argv[1]); |
| 71 | fflush(stdout); |
| 72 | argc--; |
| 73 | argv++; |
| 74 | if (ifile >= 0) |
| 75 | close(ifile); |
| 76 | } |
| 77 | exit(0); |
| 78 | } |
| 79 | |
| 80 | type(file) |
| 81 | char *file; |
| 82 | { |
| 83 | int j,nl; |
| 84 | char ch; |
| 85 | struct stat mbuf; |
| 86 | char slink[MAXPATHLEN + 1]; |
| 87 | |
| 88 | ifile = -1; |
| 89 | if (lstat(file, &mbuf) < 0) { |
| 90 | printf("%s\n", |
| 91 | (unsigned)errno < sys_nerr? sys_errlist[errno]: "Cannot stat"); |
| 92 | return; |
| 93 | } |
| 94 | switch (mbuf.st_mode & S_IFMT) { |
| 95 | |
| 96 | case S_IFLNK: |
| 97 | printf("symbolic link"); |
| 98 | j = readlink(file, slink, sizeof slink - 1); |
| 99 | if (j >= 0) { |
| 100 | slink[j] = '\0'; |
| 101 | printf(" to %s", slink); |
| 102 | } |
| 103 | printf("\n"); |
| 104 | return; |
| 105 | |
| 106 | case S_IFDIR: |
| 107 | if (mbuf.st_mode & S_ISVTX) |
| 108 | printf("append-only "); |
| 109 | printf("directory\n"); |
| 110 | return; |
| 111 | |
| 112 | case S_IFCHR: |
| 113 | case S_IFBLK: |
| 114 | printf("%s special (%d/%d)\n", |
| 115 | (mbuf.st_mode&S_IFMT) == S_IFCHR ? "character" : "block", |
| 116 | major(mbuf.st_rdev), minor(mbuf.st_rdev)); |
| 117 | return; |
| 118 | |
| 119 | case S_IFSOCK: |
| 120 | printf("socket\n"); |
| 121 | return; |
| 122 | } |
| 123 | |
| 124 | ifile = open(file, 0); |
| 125 | if(ifile < 0) { |
| 126 | printf("%s\n", |
| 127 | (unsigned)errno < sys_nerr? sys_errlist[errno]: "Cannot read"); |
| 128 | return; |
| 129 | } |
| 130 | in = read(ifile, buf, BUFSIZ); |
| 131 | if(in == 0){ |
| 132 | printf("empty\n"); |
| 133 | return; |
| 134 | } |
| 135 | switch(*(int *)buf) { |
| 136 | |
| 137 | case 0413: |
| 138 | printf("demand paged "); |
| 139 | |
| 140 | case 0410: |
| 141 | printf("pure "); |
| 142 | goto exec; |
| 143 | |
| 144 | case 0411: |
| 145 | printf("jfr or pdp-11 unix 411 executable\n"); |
| 146 | return; |
| 147 | |
| 148 | case 0407: |
| 149 | exec: |
| 150 | if (mbuf.st_mode & S_ISUID) |
| 151 | printf("set-uid "); |
| 152 | if (mbuf.st_mode & S_ISGID) |
| 153 | printf("set-gid "); |
| 154 | if (mbuf.st_mode & S_ISVTX) |
| 155 | printf("sticky "); |
| 156 | printf("executable"); |
| 157 | if(((int *)buf)[4] != 0) { |
| 158 | printf(" not stripped"); |
| 159 | if(oldo(buf)) |
| 160 | printf(" (old format symbol table)"); |
| 161 | } |
| 162 | printf("\n"); |
| 163 | return; |
| 164 | |
| 165 | case 0177555: |
| 166 | printf("very old archive\n"); |
| 167 | return; |
| 168 | |
| 169 | case 0177545: |
| 170 | printf("old archive\n"); |
| 171 | return; |
| 172 | |
| 173 | case 070707: |
| 174 | printf("cpio data\n"); |
| 175 | return; |
| 176 | } |
| 177 | |
| 178 | if (buf[0] == '#' && buf[1] == '!' && shellscript(buf+2, &mbuf)) |
| 179 | return; |
| 180 | if (buf[0] == '\037' && buf[1] == '\235') { |
| 181 | if (buf[2]&0x80) |
| 182 | printf("block "); |
| 183 | printf("compressed %d bit code data\n", buf[2]&0x1f); |
| 184 | return; |
| 185 | } |
| 186 | if(strncmp(buf, "!<arch>\n__.SYMDEF", 17) == 0 ) { |
| 187 | printf("archive random library\n"); |
| 188 | return; |
| 189 | } |
| 190 | if (strncmp(buf, "!<arch>\n", 8)==0) { |
| 191 | printf("archive\n"); |
| 192 | return; |
| 193 | } |
| 194 | if (mbuf.st_size % 512 == 0) { /* it may be a PRESS file */ |
| 195 | lseek(ifile, -512L, 2); /* last block */ |
| 196 | if (read(ifile, buf, BUFSIZ) > 0 && *(short *)buf == 12138) { |
| 197 | printf("PRESS file\n"); |
| 198 | return; |
| 199 | } |
| 200 | } |
| 201 | i = 0; |
| 202 | if(ccom() == 0)goto notc; |
| 203 | while(buf[i] == '#'){ |
| 204 | j = i; |
| 205 | while(buf[i++] != '\n'){ |
| 206 | if(i - j > 255){ |
| 207 | printf("data\n"); |
| 208 | return; |
| 209 | } |
| 210 | if(i >= in)goto notc; |
| 211 | } |
| 212 | if(ccom() == 0)goto notc; |
| 213 | } |
| 214 | check: |
| 215 | if(lookup(c) == 1){ |
| 216 | while((ch = buf[i++]) != ';' && ch != '{')if(i >= in)goto notc; |
| 217 | printf("c program text"); |
| 218 | goto outa; |
| 219 | } |
| 220 | nl = 0; |
| 221 | while(buf[i] != '('){ |
| 222 | if(buf[i] <= 0) |
| 223 | goto notas; |
| 224 | if(buf[i] == ';'){ |
| 225 | i++; |
| 226 | goto check; |
| 227 | } |
| 228 | if(buf[i++] == '\n') |
| 229 | if(nl++ > 6)goto notc; |
| 230 | if(i >= in)goto notc; |
| 231 | } |
| 232 | while(buf[i] != ')'){ |
| 233 | if(buf[i++] == '\n') |
| 234 | if(nl++ > 6)goto notc; |
| 235 | if(i >= in)goto notc; |
| 236 | } |
| 237 | while(buf[i] != '{'){ |
| 238 | if(buf[i++] == '\n') |
| 239 | if(nl++ > 6)goto notc; |
| 240 | if(i >= in)goto notc; |
| 241 | } |
| 242 | printf("c program text"); |
| 243 | goto outa; |
| 244 | notc: |
| 245 | i = 0; |
| 246 | while(buf[i] == 'c' || buf[i] == '#'){ |
| 247 | while(buf[i++] != '\n')if(i >= in)goto notfort; |
| 248 | } |
| 249 | if(lookup(fort) == 1){ |
| 250 | printf("fortran program text"); |
| 251 | goto outa; |
| 252 | } |
| 253 | notfort: |
| 254 | i=0; |
| 255 | if(ascom() == 0)goto notas; |
| 256 | j = i-1; |
| 257 | if(buf[i] == '.'){ |
| 258 | i++; |
| 259 | if(lookup(as) == 1){ |
| 260 | printf("assembler program text"); |
| 261 | goto outa; |
| 262 | } |
| 263 | else if(buf[j] == '\n' && isalpha(buf[j+2])){ |
| 264 | printf("roff, nroff, or eqn input text"); |
| 265 | goto outa; |
| 266 | } |
| 267 | } |
| 268 | while(lookup(asc) == 0){ |
| 269 | if(ascom() == 0)goto notas; |
| 270 | while(buf[i] != '\n' && buf[i++] != ':') |
| 271 | if(i >= in)goto notas; |
| 272 | while(buf[i] == '\n' || buf[i] == ' ' || buf[i] == '\t')if(i++ >= in)goto notas; |
| 273 | j = i-1; |
| 274 | if(buf[i] == '.'){ |
| 275 | i++; |
| 276 | if(lookup(as) == 1){ |
| 277 | printf("assembler program text"); |
| 278 | goto outa; |
| 279 | } |
| 280 | else if(buf[j] == '\n' && isalpha(buf[j+2])){ |
| 281 | printf("roff, nroff, or eqn input text"); |
| 282 | goto outa; |
| 283 | } |
| 284 | } |
| 285 | } |
| 286 | printf("assembler program text"); |
| 287 | goto outa; |
| 288 | notas: |
| 289 | for(i=0; i < in; i++)if(buf[i]&0200){ |
| 290 | if (buf[0]=='\100' && buf[1]=='\357') |
| 291 | printf("troff (CAT) output\n"); |
| 292 | else |
| 293 | printf("data\n"); |
| 294 | return; |
| 295 | } |
| 296 | if (mbuf.st_mode&((S_IEXEC)|(S_IEXEC>>3)|(S_IEXEC>>6))) { |
| 297 | if (mbuf.st_mode & S_ISUID) |
| 298 | printf("set-uid "); |
| 299 | if (mbuf.st_mode & S_ISGID) |
| 300 | printf("set-gid "); |
| 301 | if (mbuf.st_mode & S_ISVTX) |
| 302 | printf("sticky "); |
| 303 | if (shell(buf, in, sh)) |
| 304 | printf("shell script"); |
| 305 | else if (shell(buf, in, csh)) |
| 306 | printf("c-shell script"); |
| 307 | else |
| 308 | printf("commands text"); |
| 309 | } else if (troffint(buf, in)) |
| 310 | printf("troff intermediate output text"); |
| 311 | else if (shell(buf, in, sh)) |
| 312 | printf("shell commands"); |
| 313 | else if (shell(buf, in, csh)) |
| 314 | printf("c-shell commands"); |
| 315 | else if (english(buf, in)) |
| 316 | printf("English text"); |
| 317 | else |
| 318 | printf("ascii text"); |
| 319 | outa: |
| 320 | while(i < in) |
| 321 | if((buf[i++]&0377) > 127){ |
| 322 | printf(" with garbage\n"); |
| 323 | return; |
| 324 | } |
| 325 | /* if next few lines in then read whole file looking for nulls ... |
| 326 | while((in = read(ifile,buf,BUFSIZ)) > 0) |
| 327 | for(i = 0; i < in; i++) |
| 328 | if((buf[i]&0377) > 127){ |
| 329 | printf(" with garbage\n"); |
| 330 | return; |
| 331 | } |
| 332 | /*.... */ |
| 333 | printf("\n"); |
| 334 | } |
| 335 | |
| 336 | oldo(cp) |
| 337 | char *cp; |
| 338 | { |
| 339 | struct exec ex; |
| 340 | struct stat stb; |
| 341 | |
| 342 | ex = *(struct exec *)cp; |
| 343 | if (fstat(ifile, &stb) < 0) |
| 344 | return(0); |
| 345 | if (N_STROFF(ex)+sizeof(off_t) > stb.st_size) |
| 346 | return (1); |
| 347 | return (0); |
| 348 | } |
| 349 | |
| 350 | |
| 351 | |
| 352 | troffint(bp, n) |
| 353 | char *bp; |
| 354 | int n; |
| 355 | { |
| 356 | int k; |
| 357 | |
| 358 | i = 0; |
| 359 | for (k = 0; k < 6; k++) { |
| 360 | if (lookup(troff) == 0) |
| 361 | return(0); |
| 362 | if (lookup(troff) == 0) |
| 363 | return(0); |
| 364 | while (i < n && buf[i] != '\n') |
| 365 | i++; |
| 366 | if (i++ >= n) |
| 367 | return(0); |
| 368 | } |
| 369 | return(1); |
| 370 | } |
| 371 | lookup(tab) |
| 372 | char *tab[]; |
| 373 | { |
| 374 | char r; |
| 375 | int k,j,l; |
| 376 | while(buf[i] == ' ' || buf[i] == '\t' || buf[i] == '\n')i++; |
| 377 | for(j=0; tab[j] != 0; j++){ |
| 378 | l=0; |
| 379 | for(k=i; ((r=tab[j][l++]) == buf[k] && r != '\0');k++); |
| 380 | if(r == '\0') |
| 381 | if(buf[k] == ' ' || buf[k] == '\n' || buf[k] == '\t' |
| 382 | || buf[k] == '{' || buf[k] == '/'){ |
| 383 | i=k; |
| 384 | return(1); |
| 385 | } |
| 386 | } |
| 387 | return(0); |
| 388 | } |
| 389 | ccom(){ |
| 390 | char cc; |
| 391 | while((cc = buf[i]) == ' ' || cc == '\t' || cc == '\n')if(i++ >= in)return(0); |
| 392 | if(buf[i] == '/' && buf[i+1] == '*'){ |
| 393 | i += 2; |
| 394 | while(buf[i] != '*' || buf[i+1] != '/'){ |
| 395 | if(buf[i] == '\\')i += 2; |
| 396 | else i++; |
| 397 | if(i >= in)return(0); |
| 398 | } |
| 399 | if((i += 2) >= in)return(0); |
| 400 | } |
| 401 | if(buf[i] == '\n')if(ccom() == 0)return(0); |
| 402 | return(1); |
| 403 | } |
| 404 | ascom(){ |
| 405 | while(buf[i] == '/'){ |
| 406 | i++; |
| 407 | while(buf[i++] != '\n')if(i >= in)return(0); |
| 408 | while(buf[i] == '\n')if(i++ >= in)return(0); |
| 409 | } |
| 410 | return(1); |
| 411 | } |
| 412 | |
| 413 | english (bp, n) |
| 414 | char *bp; |
| 415 | { |
| 416 | # define NASC 128 |
| 417 | int ct[NASC], j, vow, freq, rare; |
| 418 | int badpun = 0, punct = 0; |
| 419 | if (n<50) return(0); /* no point in statistics on squibs */ |
| 420 | for(j=0; j<NASC; j++) |
| 421 | ct[j]=0; |
| 422 | for(j=0; j<n; j++) |
| 423 | { |
| 424 | if ((u_char)bp[j]<NASC) |
| 425 | ct[bp[j]|040]++; |
| 426 | switch (bp[j]) |
| 427 | { |
| 428 | case '.': |
| 429 | case ',': |
| 430 | case ')': |
| 431 | case '%': |
| 432 | case ';': |
| 433 | case ':': |
| 434 | case '?': |
| 435 | punct++; |
| 436 | if ( j < n-1 && |
| 437 | bp[j+1] != ' ' && |
| 438 | bp[j+1] != '\n') |
| 439 | badpun++; |
| 440 | } |
| 441 | } |
| 442 | if (badpun*5 > punct) |
| 443 | return(0); |
| 444 | vow = ct['a'] + ct['e'] + ct['i'] + ct['o'] + ct['u']; |
| 445 | freq = ct['e'] + ct['t'] + ct['a'] + ct['i'] + ct['o'] + ct['n']; |
| 446 | rare = ct['v'] + ct['j'] + ct['k'] + ct['q'] + ct['x'] + ct['z']; |
| 447 | if (2*ct[';'] > ct['e']) return(0); |
| 448 | if ( (ct['>']+ct['<']+ct['/'])>ct['e']) return(0); /* shell file test */ |
| 449 | return (vow*5 >= n-ct[' '] && freq >= 10*rare); |
| 450 | } |
| 451 | |
| 452 | shellscript(buf, sb) |
| 453 | char buf[]; |
| 454 | struct stat *sb; |
| 455 | { |
| 456 | register char *tp; |
| 457 | char *cp, *xp, *index(); |
| 458 | |
| 459 | cp = index(buf, '\n'); |
| 460 | if (cp == 0 || cp - buf > in) |
| 461 | return (0); |
| 462 | for (tp = buf; tp != cp && isspace(*tp); tp++) |
| 463 | if (!isascii(*tp)) |
| 464 | return (0); |
| 465 | for (xp = tp; tp != cp && !isspace(*tp); tp++) |
| 466 | if (!isascii(*tp)) |
| 467 | return (0); |
| 468 | if (tp == xp) |
| 469 | return (0); |
| 470 | if (sb->st_mode & S_ISUID) |
| 471 | printf("set-uid "); |
| 472 | if (sb->st_mode & S_ISGID) |
| 473 | printf("set-gid "); |
| 474 | if (strncmp(xp, "/bin/sh", tp-xp) == 0) |
| 475 | xp = "shell"; |
| 476 | else if (strncmp(xp, "/bin/csh", tp-xp) == 0) |
| 477 | xp = "c-shell"; |
| 478 | else |
| 479 | *tp = '\0'; |
| 480 | printf("executable %s script\n", xp); |
| 481 | return (1); |
| 482 | } |
| 483 | |
| 484 | shell(bp, n, tab) |
| 485 | char *bp; |
| 486 | int n; |
| 487 | char *tab[]; |
| 488 | { |
| 489 | |
| 490 | i = 0; |
| 491 | do { |
| 492 | if (buf[i] == '#' || buf[i] == ':') |
| 493 | while (i < n && buf[i] != '\n') |
| 494 | i++; |
| 495 | if (++i >= n) |
| 496 | break; |
| 497 | if (lookup(tab) == 1) |
| 498 | return (1); |
| 499 | } while (i < n); |
| 500 | return (0); |
| 501 | } |