xref: /original-bsd/usr.bin/file/file.c (revision 67bfb13e)
1 /*-
2  * Copyright (c) 1991 The Regents of the University of California.
3  * All rights reserved.
4  *
5  * %sccs.include.proprietary.c%
6  */
7 
8 #ifndef lint
9 char copyright[] =
10 "@(#) Copyright (c) 1991 The Regents of the University of California.\n\
11  All rights reserved.\n";
12 #endif /* not lint */
13 
14 #ifndef lint
15 static char sccsid[] = "@(#)file.c	5.2 (Berkeley) 01/11/93";
16 #endif /* not lint */
17 
18 /*
19  * file - determine type of file
20  */
21 
22 #include <sys/param.h>
23 #include <sys/stat.h>
24 #include <string.h>
25 #include <stdio.h>
26 #include <ctype.h>
27 #include <a.out.h>
28 
29 #if defined(hp300) || defined(hp800)
30 #include <hp/hpux/hpux_exec.h>
31 #endif
32 
33 extern  int errno;
34 int in;
35 int i  = 0;
36 char buf[BUFSIZ];
37 char *troff[] = {	/* new troff intermediate lang */
38 	"x","T","res","init","font","202","V0","p1",0};
39 char *fort[] = {
40 	"function","subroutine","common","dimension","block","integer",
41 	"real","data","double",0};
42 char *asc[] = {
43 	"chmk","mov","tst","clr","jmp",0};
44 char *c[] = {
45 	"int","char","float","double","struct","extern",0};
46 char *as[] = {
47 	"globl","byte","align","text","data","comm",0};
48 char *sh[] = {
49 	"fi", "elif", "esac", "done", "export",
50 	"readonly", "trap", "PATH", "HOME", 0 };
51 char *csh[] = {
52 	"alias", "breaksw", "endsw", "foreach", "limit",  "onintr",
53 	"repeat", "setenv", "source", "path", "home", 0 };
54 int	ifile;
55 
56 main(argc, argv)
57 char **argv;
58 {
59 	FILE *fl;
60 	register char *p;
61 	char ap[MAXPATHLEN + 1];
62 
63 	if (argc < 2) {
64 		fprintf(stderr, "usage: %s file ...\n", argv[0]);
65 		exit(3);
66 	}
67 
68 	if (argc>1 && argv[1][0]=='-' && argv[1][1]=='f') {
69 		if ((fl = fopen(argv[2], "r")) == NULL) {
70 			perror(argv[2]);
71 			exit(2);
72 		}
73 		while ((p = fgets(ap, sizeof ap, fl)) != NULL) {
74 			int l = strlen(p);
75 			if (l>0)
76 				p[l-1] = '\0';
77 			type(p);
78 			if (ifile>=0)
79 				close(ifile);
80 		}
81 		exit(1);
82 	}
83 	while(argc > 1) {
84 		ifile = -1;
85 		type(argv[1]);
86 		fflush(stdout);
87 		argc--;
88 		argv++;
89 		if (ifile >= 0)
90 			close(ifile);
91 	}
92 	exit(0);
93 }
94 
95 type(file)
96 char *file;
97 {
98 	int j,nl;
99 	char ch;
100 	struct stat mbuf;
101 	char slink[MAXPATHLEN + 1];
102 	struct exec *hdr;
103 #if defined(hp300) || defined(hp800)
104 	int ishpux300 = 0;
105 	int ishpux800 = 0;
106 #endif
107 
108 	if (lstat(file, &mbuf) < 0) {
109 		fprintf(stderr, "file: %s: %s\n", file, strerror(errno));
110 		return;
111 	}
112 	switch (mbuf.st_mode & S_IFMT) {
113 	case S_IFLNK:
114 		printf("%s:\tsymbolic link", file);
115 		j = readlink(file, slink, sizeof slink - 1);
116 		if (j >= 0) {
117 			slink[j] = '\0';
118 			printf(" to %s", slink);
119 		}
120 		printf("\n");
121 		return;
122 
123 	case S_IFDIR:
124 		printf("%s:\t", file);
125 		if (mbuf.st_mode & S_ISVTX)
126 			printf("append-only ");
127 		printf("directory\n");
128 		return;
129 
130 	case S_IFCHR:
131 	case S_IFBLK:
132 		printf("%s:\t%s special (%d/%d)\n", file,
133 		    (mbuf.st_mode&S_IFMT) == S_IFCHR ? "character" : "block",
134 		     major(mbuf.st_rdev), minor(mbuf.st_rdev));
135 		return;
136 
137 	case S_IFSOCK:
138 		printf("%s:\tsocket\n", file);
139 		return;
140 	}
141 
142 	ifile = open(file, 0);
143 	if (ifile < 0) {
144 		fprintf(stderr, "file: %s: %s\n", file, strerror(errno));
145 		return;
146 	}
147 	printf("%s:\t", file);
148 	in = read(ifile, buf, BUFSIZ);
149 	if (in == 0) {
150 		printf("empty\n");
151 		return;
152 	}
153 	hdr = (struct exec *) buf;
154 #ifdef MID_ZERO		/* if we have a_mid field */
155 	switch (hdr->a_mid) {
156 	case MID_SUN010:
157 		printf("SUN 68010/68020 ");
158 		break;
159 	case MID_SUN020:
160 		printf("SUN 68020 ");
161 		break;
162 	case MID_HP200:
163 		printf("HP200 ");
164 		break;
165 	case MID_HP300:
166 		printf("HP300 ");
167 		break;
168 #if defined(hp300) || defined(hp800)
169 	case MID_HPUX:
170 		printf("HP-UX series [234]00 ");
171 		ishpux300 = 1;
172 		if (hdr->a_magic == 0406) {
173 			printf("relocatable object\n");
174 			return;
175 		}
176 		break;
177 	case MID_HPUX800:
178 		printf("HP-UX series 800 ");
179 		ishpux800 = 1;
180 		if (hdr->a_magic == 0x106) {
181 			printf("relocatable object\n");
182 			return;
183 		}
184 		break;
185 #endif
186 #if BYTE_ORDER == BIG_ENDIAN
187 	case ((OMAGIC & 0xff) << 8) | (OMAGIC >> 8):
188 	case ((NMAGIC & 0xff) << 8) | (NMAGIC >> 8):
189 	case ((ZMAGIC & 0xff) << 8) | (ZMAGIC >> 8):
190 		printf("byte-swapped (VAX/386) ");
191 		hdr->a_magic = ((hdr->a_mid & 0xff) << 8) | (hdr->a_mid >> 8);
192 		break;
193 	}
194 #endif
195 #endif /* MID_ZERO, a_mid */
196 	switch (hdr->a_magic) {
197 
198 	case 0411:
199 		printf("jfr or pdp-11 unix 411 executable\n");
200 		return;
201 
202 	case ZMAGIC:
203 		printf("demand paged ");
204 		/* FALLTHROUGH */
205 
206 	case NMAGIC:
207 		printf("pure ");
208 		/* FALLTHROUGH */
209 
210 	case OMAGIC:
211 		if (mbuf.st_mode & S_ISUID)
212 			printf("set-uid ");
213 		if (mbuf.st_mode & S_ISGID)
214 			printf("set-gid ");
215 		if (mbuf.st_mode & S_ISVTX)
216 			printf("sticky ");
217 		if ((mbuf.st_mode & (S_IXUSR|S_IXGRP|S_IXOTH)) == 0 &&
218 			(hdr->a_trsize || hdr->a_drsize)) {
219 			printf("relocatable object\n");
220 			return;
221 		}
222 #if defined(hp300)
223 		if (ishpux300) {
224 			if (((int *)buf)[2] & 0x40000000)
225 				printf("dynamically-linked ");
226 		}
227 #endif
228 		printf("executable");
229 #if defined(hp300) || defined(hp800)
230 		if (ishpux300) {
231 			if (((int *)buf)[9] != 0)
232 				printf(" not stripped");
233 		} else if (ishpux800) {
234 			if (((int *)buf)[24] != 0)
235 				printf(" not stripped");
236 		} else
237 #endif
238 		if (hdr->a_syms != 0)
239 			printf(" not stripped");
240 		printf("\n");
241 		return;
242 #if defined(hp300)
243 	case 0x10e:
244 		printf("shared library, version %d\n", ((short *)buf)[2]);
245 		return;
246 #endif
247 	}
248 
249 	switch (*(int *)buf) {
250 	case 0177555:
251 		printf("very old archive\n");
252 		return;
253 
254 	case 0177545:
255 		printf("old archive\n");
256 		return;
257 
258 	case 070707:
259 		printf("cpio data\n");
260 		return;
261 	}
262 
263 	if (buf[0] == '#' && buf[1] == '!' && shellscript(buf+2, &mbuf))
264 		return;
265 	if (buf[0] == '\037' && buf[1] == '\235') {
266 		if (buf[2]&0x80)
267 			printf("block ");
268 		printf("compressed %d bit code data\n", buf[2]&0x1f);
269 		return;
270 	}
271 	if (strncmp(buf, "!<arch>\n__.SYMDEF", 17) == 0 ) {
272 		printf("archive random library\n");
273 		return;
274 	}
275 	if (strncmp(buf, "!<arch>\n", 8)==0) {
276 		printf("archive\n");
277 		return;
278 	}
279 	if (mbuf.st_size % 512 == 0) {	/* it may be a PRESS file */
280 		lseek(ifile, -512L, 2);	/* last block */
281 		if (read(ifile, buf, BUFSIZ) > 0 && *(short *)buf == 12138) {
282 			printf("PRESS file\n");
283 			return;
284 		}
285 	}
286 	i = 0;
287 	if(ccom() == 0)goto notc;
288 	while(buf[i] == '#'){
289 		j = i;
290 		while(buf[i++] != '\n'){
291 			if(i - j > 255){
292 				printf("data\n");
293 				return;
294 			}
295 			if(i >= in)goto notc;
296 		}
297 		if(ccom() == 0)goto notc;
298 	}
299 check:
300 	if(lookup(c) == 1){
301 		while((ch = buf[i++]) != ';' && ch != '{')if(i >= in)goto notc;
302 		printf("c program text");
303 		goto outa;
304 	}
305 	nl = 0;
306 	while(buf[i] != '('){
307 		if(buf[i] <= 0)
308 			goto notas;
309 		if(buf[i] == ';'){
310 			i++;
311 			goto check;
312 		}
313 		if(buf[i++] == '\n')
314 			if(nl++ > 6)goto notc;
315 		if(i >= in)goto notc;
316 	}
317 	while(buf[i] != ')'){
318 		if(buf[i++] == '\n')
319 			if(nl++ > 6)goto notc;
320 		if(i >= in)goto notc;
321 	}
322 	while(buf[i] != '{'){
323 		if(buf[i++] == '\n')
324 			if(nl++ > 6)goto notc;
325 		if(i >= in)goto notc;
326 	}
327 	printf("c program text");
328 	goto outa;
329 notc:
330 	i = 0;
331 	while(buf[i] == 'c' || buf[i] == '#'){
332 		while(buf[i++] != '\n')if(i >= in)goto notfort;
333 	}
334 	if(lookup(fort) == 1){
335 		printf("fortran program text");
336 		goto outa;
337 	}
338 notfort:
339 	i=0;
340 	if(ascom() == 0)goto notas;
341 	j = i-1;
342 	if(buf[i] == '.'){
343 		i++;
344 		if(lookup(as) == 1){
345 			printf("assembler program text");
346 			goto outa;
347 		}
348 		else if(buf[j] == '\n' && isalpha(buf[j+2])){
349 			printf("roff, nroff, or eqn input text");
350 			goto outa;
351 		}
352 	}
353 	while(lookup(asc) == 0){
354 		if(ascom() == 0)goto notas;
355 		while(buf[i] != '\n' && buf[i++] != ':')
356 			if(i >= in)goto notas;
357 		while(buf[i] == '\n' || buf[i] == ' ' || buf[i] == '\t')if(i++ >= in)goto notas;
358 		j = i-1;
359 		if(buf[i] == '.'){
360 			i++;
361 			if(lookup(as) == 1){
362 				printf("assembler program text");
363 				goto outa;
364 			}
365 			else if(buf[j] == '\n' && isalpha(buf[j+2])){
366 				printf("roff, nroff, or eqn input text");
367 				goto outa;
368 			}
369 		}
370 	}
371 	printf("assembler program text");
372 	goto outa;
373 notas:
374 	for(i=0; i < in; i++)if(buf[i]&0200){
375 		if (buf[0]=='\100' && buf[1]=='\357')
376 			printf("troff (CAT) output\n");
377 		else
378 			printf("data\n");
379 		return;
380 	}
381 	if (mbuf.st_mode&((S_IEXEC)|(S_IEXEC>>3)|(S_IEXEC>>6))) {
382 		if (mbuf.st_mode & S_ISUID)
383 			printf("set-uid ");
384 		if (mbuf.st_mode & S_ISGID)
385 			printf("set-gid ");
386 		if (mbuf.st_mode & S_ISVTX)
387 			printf("sticky ");
388 		if (shell(buf, in, sh))
389 			printf("shell script");
390 		else if (shell(buf, in, csh))
391 			printf("c-shell script");
392 		else
393 			printf("commands text");
394 	} else if (troffint(buf, in))
395 		printf("troff intermediate output text");
396 	else if (shell(buf, in, sh))
397 		printf("shell commands");
398 	else if (shell(buf, in, csh))
399 		printf("c-shell commands");
400 	else if (english(buf, in))
401 		printf("English text");
402 	else
403 		printf("ascii text");
404 outa:
405 	while(i < in)
406 		if((buf[i++]&0377) > 127){
407 			printf(" with garbage\n");
408 			return;
409 		}
410 	/* if next few lines in then read whole file looking for nulls ...
411 		while((in = read(ifile,buf,BUFSIZ)) > 0)
412 			for(i = 0; i < in; i++)
413 				if((buf[i]&0377) > 127){
414 					printf(" with garbage\n");
415 					return;
416 				}
417 		/*.... */
418 	printf("\n");
419 }
420 
421 troffint(bp, n)
422 char *bp;
423 int n;
424 {
425 	int k;
426 
427 	i = 0;
428 	for (k = 0; k < 6; k++) {
429 		if (lookup(troff) == 0)
430 			return(0);
431 		if (lookup(troff) == 0)
432 			return(0);
433 		while (i < n && buf[i] != '\n')
434 			i++;
435 		if (i++ >= n)
436 			return(0);
437 	}
438 	return(1);
439 }
440 lookup(tab)
441 char *tab[];
442 {
443 	char r;
444 	int k,j,l;
445 	while(buf[i] == ' ' || buf[i] == '\t' || buf[i] == '\n')i++;
446 	for(j=0; tab[j] != 0; j++){
447 		l=0;
448 		for(k=i; ((r=tab[j][l++]) == buf[k] && r != '\0');k++);
449 		if(r == '\0')
450 			if(buf[k] == ' ' || buf[k] == '\n' || buf[k] == '\t'
451 			    || buf[k] == '{' || buf[k] == '/'){
452 				i=k;
453 				return(1);
454 			}
455 	}
456 	return(0);
457 }
458 ccom(){
459 	char cc;
460 	while((cc = buf[i]) == ' ' || cc == '\t' || cc == '\n')if(i++ >= in)return(0);
461 	if(buf[i] == '/' && buf[i+1] == '*'){
462 		i += 2;
463 		while(buf[i] != '*' || buf[i+1] != '/'){
464 			if(buf[i] == '\\')i += 2;
465 			else i++;
466 			if(i >= in)return(0);
467 		}
468 		if((i += 2) >= in)return(0);
469 	}
470 	if(buf[i] == '\n')if(ccom() == 0)return(0);
471 	return(1);
472 }
473 ascom(){
474 	while(buf[i] == '/'){
475 		i++;
476 		while(buf[i++] != '\n')if(i >= in)return(0);
477 		while(buf[i] == '\n')if(i++ >= in)return(0);
478 	}
479 	return(1);
480 }
481 
482 english (bp, n)
483 char *bp;
484 {
485 # define NASC 128
486 	int ct[NASC], j, vow, freq, rare;
487 	int badpun = 0, punct = 0;
488 	if (n<50) return(0); /* no point in statistics on squibs */
489 	for(j=0; j<NASC; j++)
490 		ct[j]=0;
491 	for(j=0; j<n; j++)
492 	{
493 		if ((u_char)bp[j]<NASC)
494 			ct[bp[j]|040]++;
495 		switch (bp[j])
496 		{
497 		case '.':
498 		case ',':
499 		case ')':
500 		case '%':
501 		case ';':
502 		case ':':
503 		case '?':
504 			punct++;
505 			if ( j < n-1 &&
506 			    bp[j+1] != ' ' &&
507 			    bp[j+1] != '\n')
508 				badpun++;
509 		}
510 	}
511 	if (badpun*5 > punct)
512 		return(0);
513 	vow = ct['a'] + ct['e'] + ct['i'] + ct['o'] + ct['u'];
514 	freq = ct['e'] + ct['t'] + ct['a'] + ct['i'] + ct['o'] + ct['n'];
515 	rare = ct['v'] + ct['j'] + ct['k'] + ct['q'] + ct['x'] + ct['z'];
516 	if (2*ct[';'] > ct['e']) return(0);
517 	if ( (ct['>']+ct['<']+ct['/'])>ct['e']) return(0); /* shell file test */
518 	return (vow*5 >= n-ct[' '] && freq >= 10*rare);
519 }
520 
521 shellscript(buf, sb)
522 	char buf[];
523 	struct stat *sb;
524 {
525 	register char *tp;
526 	char *cp, *xp, *index();
527 
528 	cp = index(buf, '\n');
529 	if (cp == 0 || cp - buf > in)
530 		return (0);
531 	for (tp = buf; tp != cp && isspace(*tp); tp++)
532 		if (!isascii(*tp))
533 			return (0);
534 	for (xp = tp; tp != cp && !isspace(*tp); tp++)
535 		if (!isascii(*tp))
536 			return (0);
537 	if (tp == xp)
538 		return (0);
539 	if (sb->st_mode & S_ISUID)
540 		printf("set-uid ");
541 	if (sb->st_mode & S_ISGID)
542 		printf("set-gid ");
543 	if (strncmp(xp, "/bin/sh", tp-xp) == 0)
544 		xp = "shell";
545 	else if (strncmp(xp, "/bin/csh", tp-xp) == 0)
546 		xp = "c-shell";
547 	else
548 		*tp = '\0';
549 	printf("executable %s script\n", xp);
550 	return (1);
551 }
552 
553 shell(bp, n, tab)
554 	char *bp;
555 	int n;
556 	char *tab[];
557 {
558 
559 	i = 0;
560 	do {
561 		if (buf[i] == '#' || buf[i] == ':')
562 			while (i < n && buf[i] != '\n')
563 				i++;
564 		if (++i >= n)
565 			break;
566 		if (lookup(tab) == 1)
567 			return (1);
568 	} while (i < n);
569 	return (0);
570 }
571