2.11BSD#431 - /usr/src/usr.bin/file.c

   1: #if !defined(lint) && defined(DOSCCS)
   2: static  char sccsid[] = "@(#)file.c	4.12.3 (2.11BSD) 1997/7/29";
   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: #include <errno.h>
  14: #include <string.h>
  15: 
  16: int in;
  17: int i  = 0;
  18: char buf[BUFSIZ];
  19: char *troff[] = {   /* new troff intermediate lang */
  20:     "x","T","res","init","font","202","V0","p1",0};
  21: char *fort[] = {
  22:     "function","subroutine","common","dimension","block","integer",
  23:     "real","data","double",0};
  24: char *asc[] = {
  25:     "sys","mov","tst","clr","jmp",0};
  26: char *c[] = {
  27:     "int","char","float","double","struct","extern",0};
  28: char *as[] = {
  29:     "globl","byte","align","text","data","comm",0};
  30: char *sh[] = {
  31:     "fi", "elif", "esac", "done", "export",
  32:     "readonly", "trap", "PATH", "HOME", 0 };
  33: char *csh[] = {
  34:     "alias", "breaksw", "endsw", "foreach", "limit",  "onintr",
  35:     "repeat", "setenv", "source", "path", "home", 0 };
  36: int ifile;
  37: 
  38: main(argc, argv)
  39: char **argv;
  40: {
  41:     register FILE *fl;
  42:     register char *p;
  43:     char ap[MAXPATHLEN + 1];
  44: 
  45:     if (argc < 2) {
  46:         fprintf(stderr, "usage: %s file ...\n", argv[0]);
  47:         exit(3);
  48:     }
  49: 
  50:     if (argc>1 && argv[1][0]=='-' && argv[1][1]=='f') {
  51:         if ((fl = fopen(argv[2], "r")) == NULL) {
  52:             perror(argv[2]);
  53:             exit(2);
  54:         }
  55:         while ((p = fgets(ap, sizeof ap, fl)) != NULL) {
  56:             int l = strlen(p);
  57:             if (l>0)
  58:                 p[l-1] = '\0';
  59:             printf("%s:	", p);
  60:             type(p);
  61:             if (ifile>=0)
  62:                 close(ifile);
  63:         }
  64:         exit(1);
  65:     }
  66:     while(argc > 1) {
  67:         printf("%s:	", argv[1]);
  68:         type(argv[1]);
  69:         fflush(stdout);
  70:         argc--;
  71:         argv++;
  72:         if (ifile >= 0)
  73:             close(ifile);
  74:     }
  75:     exit(0);
  76: }
  77: 
  78: type(file)
  79: char *file;
  80: {
  81:     register int j;
  82:     int nl;
  83:     char ch;
  84:     struct stat mbuf;
  85:     char slink[MAXPATHLEN + 1];
  86: 
  87:     ifile = -1;
  88:     if (lstat(file, &mbuf) < 0) {
  89:         printf("%s\n", strerror(errno));
  90:         return;
  91:     }
  92:     switch (mbuf.st_mode & S_IFMT) {
  93: 
  94:     case S_IFLNK:
  95:         printf("symbolic link");
  96:         j = readlink(file, slink, sizeof slink - 1);
  97:         if (j >= 0) {
  98:             slink[j] = '\0';
  99:             printf(" to %s", slink);
 100:         }
 101:         printf("\n");
 102:         return;
 103: 
 104:     case S_IFDIR:
 105:         if (mbuf.st_mode & S_ISVTX)
 106:             printf("append-only ");
 107:         printf("directory\n");
 108:         return;
 109: 
 110:     case S_IFCHR:
 111:     case S_IFBLK:
 112:         printf("%s special (%d/%d)\n",
 113:             (mbuf.st_mode&S_IFMT) == S_IFCHR ? "character" : "block",
 114:              major(mbuf.st_rdev), minor(mbuf.st_rdev));
 115:         return;
 116: 
 117:     case S_IFSOCK:
 118:         printf("socket\n");
 119:         return;
 120:     }
 121: 
 122:     ifile = open(file, 0);
 123:     if(ifile < 0) {
 124:         printf("%s\n", strerror(errno));
 125:         return;
 126:     }
 127:     in = read(ifile, buf, BUFSIZ);
 128:     if(in == 0){
 129:         printf("empty\n");
 130:         return;
 131:     }
 132:     switch(*(int *)buf) {
 133: 
 134:     case 0413:
 135:         printf("demand paged ");
 136: 
 137:     case 0410:
 138:         printf("pure ");
 139:         goto exec;
 140: 
 141:     case 0411:
 142:         printf("separate ");
 143:         goto exec;
 144: 
 145:     case A_MAGIC4:
 146:         printf("replacement text ");
 147:         goto exec;
 148: 
 149:     case A_MAGIC5:
 150:         printf("overlaid pure ");
 151:         goto exec;
 152: 
 153:     case A_MAGIC6:
 154:         printf("overlaid separate ");
 155:         goto exec;
 156: 
 157:     case 0407:
 158: exec:
 159:         if (mbuf.st_mode & S_ISUID)
 160:             printf("set-uid ");
 161:         if (mbuf.st_mode & S_ISGID)
 162:             printf("set-gid ");
 163:         if (mbuf.st_mode & S_ISVTX)
 164:             printf("sticky ");
 165:         printf("executable");
 166:         if(((int *)buf)[4] != 0)
 167:             printf(" not stripped");
 168:         if (((int *)buf)[1] == 0 && ((int *)buf)[2] != 0 &&
 169:             ((int *)buf)[3] == 0)
 170:             printf(" (likely vax)");
 171:         printf("\n");
 172:         return;
 173: 
 174:     case 0177555:
 175:         printf("very old archive\n");
 176:         return;
 177: 
 178:     case 0177545:
 179:         printf("old archive\n");
 180:         return;
 181: 
 182:     case 070707:
 183:         printf("cpio data\n");
 184:         return;
 185:     }
 186: 
 187:     if (buf[0] == '#' && buf[1] == '!' && shellscript(buf+2, &mbuf))
 188:         return;
 189:     if (buf[0] == '\037' && buf[1] == '\235') {
 190:         if (buf[2]&0x80)
 191:             printf("block ");
 192:         printf("compressed %d bit code data\n", buf[2]&0x1f);
 193:         return;
 194:     }
 195:     if(strncmp(buf, "!<arch>\n__.SYMDEF", 17) == 0 ) {
 196:         printf("archive random library\n");
 197:         return;
 198:     }
 199:     if (strncmp(buf, "!<arch>\n", 8)==0) {
 200:         printf("archive\n");
 201:         return;
 202:     }
 203:     if (mbuf.st_size % 512 == 0) {  /* it may be a PRESS file */
 204:         lseek(ifile, -512L, 2); /* last block */
 205:         if (read(ifile, buf, BUFSIZ) > 0 && *(short *)buf == 12138) {
 206:             printf("PRESS file\n");
 207:             return;
 208:         }
 209:     }
 210:     i = 0;
 211:     if(ccom() == 0)goto notc;
 212:     while(buf[i] == '#'){
 213:         j = i;
 214:         while(buf[i++] != '\n'){
 215:             if(i - j > 255){
 216:                 printf("data\n");
 217:                 return;
 218:             }
 219:             if(i >= in)goto notc;
 220:         }
 221:         if(ccom() == 0)goto notc;
 222:     }
 223: check:
 224:     if(lookup(c) == 1){
 225:         while((ch = buf[i++]) != ';' && ch != '{')if(i >= in)goto notc;
 226:         printf("c program text");
 227:         goto outa;
 228:     }
 229:     nl = 0;
 230:     while(buf[i] != '('){
 231:         if(buf[i] <= 0)
 232:             goto notas;
 233:         if(buf[i] == ';'){
 234:             i++;
 235:             goto check;
 236:         }
 237:         if(buf[i++] == '\n')
 238:             if(nl++ > 6)goto notc;
 239:         if(i >= in)goto notc;
 240:     }
 241:     while(buf[i] != ')'){
 242:         if(buf[i++] == '\n')
 243:             if(nl++ > 6)goto notc;
 244:         if(i >= in)goto notc;
 245:     }
 246:     while(buf[i] != '{'){
 247:         if(buf[i++] == '\n')
 248:             if(nl++ > 6)goto notc;
 249:         if(i >= in)goto notc;
 250:     }
 251:     printf("c program text");
 252:     goto outa;
 253: notc:
 254:     i = 0;
 255:     while(buf[i] == 'c' || buf[i] == '#'){
 256:         while(buf[i++] != '\n')if(i >= in)goto notfort;
 257:     }
 258:     if(lookup(fort) == 1){
 259:         printf("fortran program text");
 260:         goto outa;
 261:     }
 262: notfort:
 263:     i=0;
 264:     if(ascom() == 0)goto notas;
 265:     j = i-1;
 266:     if(buf[i] == '.'){
 267:         i++;
 268:         if(lookup(as) == 1){
 269:             printf("assembler program text");
 270:             goto outa;
 271:         }
 272:         else if(buf[j] == '\n' && isalpha(buf[j+2])){
 273:             printf("roff, nroff, or eqn input text");
 274:             goto outa;
 275:         }
 276:     }
 277:     while(lookup(asc) == 0){
 278:         if(ascom() == 0)goto notas;
 279:         while(buf[i] != '\n' && buf[i++] != ':')
 280:             if(i >= in)goto notas;
 281:         while(buf[i] == '\n' || buf[i] == ' ' || buf[i] == '\t')if(i++ >= in)goto notas;
 282:         j = i-1;
 283:         if(buf[i] == '.'){
 284:             i++;
 285:             if(lookup(as) == 1){
 286:                 printf("assembler program text");
 287:                 goto outa;
 288:             }
 289:             else if(buf[j] == '\n' && isalpha(buf[j+2])){
 290:                 printf("roff, nroff, or eqn input text");
 291:                 goto outa;
 292:             }
 293:         }
 294:     }
 295:     printf("assembler program text");
 296:     goto outa;
 297: notas:
 298:     for(i=0; i < in; i++)if(buf[i]&0200){
 299:         if (buf[0]=='\100' && buf[1]=='\357')
 300:             printf("troff (CAT) output\n");
 301:         else
 302:             printf("data\n");
 303:         return;
 304:     }
 305:     if (mbuf.st_mode&((S_IEXEC)|(S_IEXEC>>3)|(S_IEXEC>>6))) {
 306:         if (mbuf.st_mode & S_ISUID)
 307:             printf("set-uid ");
 308:         if (mbuf.st_mode & S_ISGID)
 309:             printf("set-gid ");
 310:         if (mbuf.st_mode & S_ISVTX)
 311:             printf("sticky ");
 312:         if (shell(buf, in, sh))
 313:             printf("shell script");
 314:         else if (shell(buf, in, csh))
 315:             printf("c-shell script");
 316:         else
 317:             printf("commands text");
 318:     } else if (troffint(buf, in))
 319:         printf("troff intermediate output text");
 320:     else if (shell(buf, in, sh))
 321:         printf("shell commands");
 322:     else if (shell(buf, in, csh))
 323:         printf("c-shell commands");
 324:     else if (english(buf, in))
 325:         printf("English text");
 326:     else
 327:         printf("ascii text");
 328: outa:
 329:     while(i < in)
 330:         if((buf[i++]&0377) > 127){
 331:             printf(" with garbage\n");
 332:             return;
 333:         }
 334:     /* if next few lines in then read whole file looking for nulls ...
 335: 		while((in = read(ifile,buf,BUFSIZ)) > 0)
 336: 			for(i = 0; i < in; i++)
 337: 				if((buf[i]&0377) > 127){
 338: 					printf(" with garbage\n");
 339: 					return;
 340: 				}
 341: 		/*.... */
 342:     printf("\n");
 343: }
 344: 
 345: troffint(bp, n)
 346: char *bp;
 347: int n;
 348: {
 349:     int k;
 350: 
 351:     i = 0;
 352:     for (k = 0; k < 6; k++) {
 353:         if (lookup(troff) == 0)
 354:             return(0);
 355:         if (lookup(troff) == 0)
 356:             return(0);
 357:         while (i < n && buf[i] != '\n')
 358:             i++;
 359:         if (i++ >= n)
 360:             return(0);
 361:     }
 362:     return(1);
 363: }
 364: lookup(tab)
 365: char *tab[];
 366: {
 367:     char r;
 368:     int k,j,l;
 369:     while(buf[i] == ' ' || buf[i] == '\t' || buf[i] == '\n')i++;
 370:     for(j=0; tab[j] != 0; j++){
 371:         l=0;
 372:         for(k=i; ((r=tab[j][l++]) == buf[k] && r != '\0');k++);
 373:         if(r == '\0')
 374:             if(buf[k] == ' ' || buf[k] == '\n' || buf[k] == '\t'
 375:                 || buf[k] == '{' || buf[k] == '/'){
 376:                 i=k;
 377:                 return(1);
 378:             }
 379:     }
 380:     return(0);
 381: }
 382: ccom(){
 383:     char cc;
 384:     while((cc = buf[i]) == ' ' || cc == '\t' || cc == '\n')if(i++ >= in)return(0);
 385:     if(buf[i] == '/' && buf[i+1] == '*'){
 386:         i += 2;
 387:         while(buf[i] != '*' || buf[i+1] != '/'){
 388:             if(buf[i] == '\\')i += 2;
 389:             else i++;
 390:             if(i >= in)return(0);
 391:         }
 392:         if((i += 2) >= in)return(0);
 393:     }
 394:     if(buf[i] == '\n')if(ccom() == 0)return(0);
 395:     return(1);
 396: }
 397: ascom(){
 398:     while(buf[i] == '/'){
 399:         i++;
 400:         while(buf[i++] != '\n')if(i >= in)return(0);
 401:         while(buf[i] == '\n')if(i++ >= in)return(0);
 402:     }
 403:     return(1);
 404: }
 405: 
 406: english (bp, n)
 407: char *bp;
 408: {
 409: # define NASC 128
 410:     int ct[NASC], j, vow, freq, rare;
 411:     int badpun = 0, punct = 0;
 412:     if (n<50) return(0); /* no point in statistics on squibs */
 413:     for(j=0; j<NASC; j++)
 414:         ct[j]=0;
 415:     for(j=0; j<n; j++)
 416:     {
 417:         if (bp[j]<NASC)
 418:             ct[bp[j]|040]++;
 419:         switch (bp[j])
 420:         {
 421:         case '.':
 422:         case ',':
 423:         case ')':
 424:         case '%':
 425:         case ';':
 426:         case ':':
 427:         case '?':
 428:             punct++;
 429:             if ( j < n-1 &&
 430:                 bp[j+1] != ' ' &&
 431:                 bp[j+1] != '\n')
 432:                 badpun++;
 433:         }
 434:     }
 435:     if (badpun*5 > punct)
 436:         return(0);
 437:     vow = ct['a'] + ct['e'] + ct['i'] + ct['o'] + ct['u'];
 438:     freq = ct['e'] + ct['t'] + ct['a'] + ct['i'] + ct['o'] + ct['n'];
 439:     rare = ct['v'] + ct['j'] + ct['k'] + ct['q'] + ct['x'] + ct['z'];
 440:     if (2*ct[';'] > ct['e']) return(0);
 441:     if ( (ct['>']+ct['<']+ct['/'])>ct['e']) return(0); /* shell file test */
 442:     return (vow*5 >= n-ct[' '] && freq >= 10*rare);
 443: }
 444: 
 445: shellscript(buf, sb)
 446:     char buf[];
 447:     struct stat *sb;
 448: {
 449:     register char *tp;
 450:     char *cp, *xp, *index();
 451: 
 452:     cp = index(buf, '\n');
 453:     if (cp == 0 || cp - buf > in)
 454:         return (0);
 455:     for (tp = buf; tp != cp && isspace(*tp); tp++)
 456:         if (!isascii(*tp))
 457:             return (0);
 458:     for (xp = tp; tp != cp && !isspace(*tp); tp++)
 459:         if (!isascii(*tp))
 460:             return (0);
 461:     if (tp == xp)
 462:         return (0);
 463:     if (sb->st_mode & S_ISUID)
 464:         printf("set-uid ");
 465:     if (sb->st_mode & S_ISGID)
 466:         printf("set-gid ");
 467:     if (strncmp(xp, "/bin/sh", tp-xp) == 0)
 468:         xp = "shell";
 469:     else if (strncmp(xp, "/bin/csh", tp-xp) == 0)
 470:         xp = "c-shell";
 471:     else
 472:         *tp = '\0';
 473:     printf("executable %s script\n", xp);
 474:     return (1);
 475: }
 476: 
 477: shell(bp, n, tab)
 478:     char *bp;
 479:     int n;
 480:     char *tab[];
 481: {
 482: 
 483:     i = 0;
 484:     do {
 485:         if (buf[i] == '#' || buf[i] == ':')
 486:             while (i < n && buf[i] != '\n')
 487:                 i++;
 488:         if (++i >= n)
 489:             break;
 490:         if (lookup(tab) == 1)
 491:             return (1);
 492:     } while (i < n);
 493:     return (0);
 494: }

Defined functions

Defined variables

Defined macros