xref: /original-bsd/usr.bin/file/file.c (revision a95f03a8)
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	5.1 (Berkeley) 06/17/92";
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 		if (((struct hpux_exec *)buf)->ha_version == BSDVNUM)
171 			printf("BSD generated ");
172 		printf("HP-UX series 200/300 ");
173 		ishpux300 = 1;
174 		if (hdr->a_magic == 0406) {
175 			printf("relocatable object\n");
176 			return;
177 		}
178 		break;
179 	case MID_HPUX800:
180 		printf("HP-UX series 800 ");
181 		ishpux800 = 1;
182 		if (hdr->a_magic == 0x106) {
183 			printf("relocatable object\n");
184 			return;
185 		}
186 		break;
187 #endif
188 #if BYTE_ORDER == BIG_ENDIAN
189 	case ((OMAGIC & 0xff) << 8) | (OMAGIC >> 8):
190 	case ((NMAGIC & 0xff) << 8) | (NMAGIC >> 8):
191 	case ((ZMAGIC & 0xff) << 8) | (ZMAGIC >> 8):
192 		printf("byte-swapped (VAX/386) ");
193 		hdr->a_magic = ((hdr->a_mid & 0xff) << 8) | (hdr->a_mid >> 8);
194 		break;
195 	}
196 #endif
197 #endif /* MID_ZERO, a_mid */
198 	switch (hdr->a_magic) {
199 
200 	case 0411:
201 		printf("jfr or pdp-11 unix 411 executable\n");
202 		return;
203 
204 	case ZMAGIC:
205 		printf("demand paged ");
206 		/* FALLTHROUGH */
207 
208 	case NMAGIC:
209 		printf("pure ");
210 		/* FALLTHROUGH */
211 
212 	case OMAGIC:
213 		if (mbuf.st_mode & S_ISUID)
214 			printf("set-uid ");
215 		if (mbuf.st_mode & S_ISGID)
216 			printf("set-gid ");
217 		if (mbuf.st_mode & S_ISVTX)
218 			printf("sticky ");
219 		if ((mbuf.st_mode & (S_IXUSR|S_IXGRP|S_IXOTH)) == 0 &&
220 			(hdr->a_trsize || hdr->a_drsize)) {
221 			printf("relocatable object\n");
222 			return;
223 		}
224 		printf("executable");
225 #if defined(hp300) || defined(hp800)
226 		if (ishpux300) {
227 			if (((int *)buf)[9] != 0)
228 				printf(" not stripped");
229 		} else if (ishpux800) {
230 			if (((int *)buf)[24] != 0)
231 				printf(" not stripped");
232 		} else
233 #endif
234 		if (hdr->a_syms != 0)
235 			printf(" not stripped");
236 		printf("\n");
237 		return;
238 	}
239 
240 	switch (*(int *)buf) {
241 	case 0177555:
242 		printf("very old archive\n");
243 		return;
244 
245 	case 0177545:
246 		printf("old archive\n");
247 		return;
248 
249 	case 070707:
250 		printf("cpio data\n");
251 		return;
252 	}
253 
254 	if (buf[0] == '#' && buf[1] == '!' && shellscript(buf+2, &mbuf))
255 		return;
256 	if (buf[0] == '\037' && buf[1] == '\235') {
257 		if (buf[2]&0x80)
258 			printf("block ");
259 		printf("compressed %d bit code data\n", buf[2]&0x1f);
260 		return;
261 	}
262 	if (strncmp(buf, "!<arch>\n__.SYMDEF", 17) == 0 ) {
263 		printf("archive random library\n");
264 		return;
265 	}
266 	if (strncmp(buf, "!<arch>\n", 8)==0) {
267 		printf("archive\n");
268 		return;
269 	}
270 	if (mbuf.st_size % 512 == 0) {	/* it may be a PRESS file */
271 		lseek(ifile, -512L, 2);	/* last block */
272 		if (read(ifile, buf, BUFSIZ) > 0 && *(short *)buf == 12138) {
273 			printf("PRESS file\n");
274 			return;
275 		}
276 	}
277 	i = 0;
278 	if(ccom() == 0)goto notc;
279 	while(buf[i] == '#'){
280 		j = i;
281 		while(buf[i++] != '\n'){
282 			if(i - j > 255){
283 				printf("data\n");
284 				return;
285 			}
286 			if(i >= in)goto notc;
287 		}
288 		if(ccom() == 0)goto notc;
289 	}
290 check:
291 	if(lookup(c) == 1){
292 		while((ch = buf[i++]) != ';' && ch != '{')if(i >= in)goto notc;
293 		printf("c program text");
294 		goto outa;
295 	}
296 	nl = 0;
297 	while(buf[i] != '('){
298 		if(buf[i] <= 0)
299 			goto notas;
300 		if(buf[i] == ';'){
301 			i++;
302 			goto check;
303 		}
304 		if(buf[i++] == '\n')
305 			if(nl++ > 6)goto notc;
306 		if(i >= in)goto notc;
307 	}
308 	while(buf[i] != ')'){
309 		if(buf[i++] == '\n')
310 			if(nl++ > 6)goto notc;
311 		if(i >= in)goto notc;
312 	}
313 	while(buf[i] != '{'){
314 		if(buf[i++] == '\n')
315 			if(nl++ > 6)goto notc;
316 		if(i >= in)goto notc;
317 	}
318 	printf("c program text");
319 	goto outa;
320 notc:
321 	i = 0;
322 	while(buf[i] == 'c' || buf[i] == '#'){
323 		while(buf[i++] != '\n')if(i >= in)goto notfort;
324 	}
325 	if(lookup(fort) == 1){
326 		printf("fortran program text");
327 		goto outa;
328 	}
329 notfort:
330 	i=0;
331 	if(ascom() == 0)goto notas;
332 	j = i-1;
333 	if(buf[i] == '.'){
334 		i++;
335 		if(lookup(as) == 1){
336 			printf("assembler program text");
337 			goto outa;
338 		}
339 		else if(buf[j] == '\n' && isalpha(buf[j+2])){
340 			printf("roff, nroff, or eqn input text");
341 			goto outa;
342 		}
343 	}
344 	while(lookup(asc) == 0){
345 		if(ascom() == 0)goto notas;
346 		while(buf[i] != '\n' && buf[i++] != ':')
347 			if(i >= in)goto notas;
348 		while(buf[i] == '\n' || buf[i] == ' ' || buf[i] == '\t')if(i++ >= in)goto notas;
349 		j = i-1;
350 		if(buf[i] == '.'){
351 			i++;
352 			if(lookup(as) == 1){
353 				printf("assembler program text");
354 				goto outa;
355 			}
356 			else if(buf[j] == '\n' && isalpha(buf[j+2])){
357 				printf("roff, nroff, or eqn input text");
358 				goto outa;
359 			}
360 		}
361 	}
362 	printf("assembler program text");
363 	goto outa;
364 notas:
365 	for(i=0; i < in; i++)if(buf[i]&0200){
366 		if (buf[0]=='\100' && buf[1]=='\357')
367 			printf("troff (CAT) output\n");
368 		else
369 			printf("data\n");
370 		return;
371 	}
372 	if (mbuf.st_mode&((S_IEXEC)|(S_IEXEC>>3)|(S_IEXEC>>6))) {
373 		if (mbuf.st_mode & S_ISUID)
374 			printf("set-uid ");
375 		if (mbuf.st_mode & S_ISGID)
376 			printf("set-gid ");
377 		if (mbuf.st_mode & S_ISVTX)
378 			printf("sticky ");
379 		if (shell(buf, in, sh))
380 			printf("shell script");
381 		else if (shell(buf, in, csh))
382 			printf("c-shell script");
383 		else
384 			printf("commands text");
385 	} else if (troffint(buf, in))
386 		printf("troff intermediate output text");
387 	else if (shell(buf, in, sh))
388 		printf("shell commands");
389 	else if (shell(buf, in, csh))
390 		printf("c-shell commands");
391 	else if (english(buf, in))
392 		printf("English text");
393 	else
394 		printf("ascii text");
395 outa:
396 	while(i < in)
397 		if((buf[i++]&0377) > 127){
398 			printf(" with garbage\n");
399 			return;
400 		}
401 	/* if next few lines in then read whole file looking for nulls ...
402 		while((in = read(ifile,buf,BUFSIZ)) > 0)
403 			for(i = 0; i < in; i++)
404 				if((buf[i]&0377) > 127){
405 					printf(" with garbage\n");
406 					return;
407 				}
408 		/*.... */
409 	printf("\n");
410 }
411 
412 troffint(bp, n)
413 char *bp;
414 int n;
415 {
416 	int k;
417 
418 	i = 0;
419 	for (k = 0; k < 6; k++) {
420 		if (lookup(troff) == 0)
421 			return(0);
422 		if (lookup(troff) == 0)
423 			return(0);
424 		while (i < n && buf[i] != '\n')
425 			i++;
426 		if (i++ >= n)
427 			return(0);
428 	}
429 	return(1);
430 }
431 lookup(tab)
432 char *tab[];
433 {
434 	char r;
435 	int k,j,l;
436 	while(buf[i] == ' ' || buf[i] == '\t' || buf[i] == '\n')i++;
437 	for(j=0; tab[j] != 0; j++){
438 		l=0;
439 		for(k=i; ((r=tab[j][l++]) == buf[k] && r != '\0');k++);
440 		if(r == '\0')
441 			if(buf[k] == ' ' || buf[k] == '\n' || buf[k] == '\t'
442 			    || buf[k] == '{' || buf[k] == '/'){
443 				i=k;
444 				return(1);
445 			}
446 	}
447 	return(0);
448 }
449 ccom(){
450 	char cc;
451 	while((cc = buf[i]) == ' ' || cc == '\t' || cc == '\n')if(i++ >= in)return(0);
452 	if(buf[i] == '/' && buf[i+1] == '*'){
453 		i += 2;
454 		while(buf[i] != '*' || buf[i+1] != '/'){
455 			if(buf[i] == '\\')i += 2;
456 			else i++;
457 			if(i >= in)return(0);
458 		}
459 		if((i += 2) >= in)return(0);
460 	}
461 	if(buf[i] == '\n')if(ccom() == 0)return(0);
462 	return(1);
463 }
464 ascom(){
465 	while(buf[i] == '/'){
466 		i++;
467 		while(buf[i++] != '\n')if(i >= in)return(0);
468 		while(buf[i] == '\n')if(i++ >= in)return(0);
469 	}
470 	return(1);
471 }
472 
473 english (bp, n)
474 char *bp;
475 {
476 # define NASC 128
477 	int ct[NASC], j, vow, freq, rare;
478 	int badpun = 0, punct = 0;
479 	if (n<50) return(0); /* no point in statistics on squibs */
480 	for(j=0; j<NASC; j++)
481 		ct[j]=0;
482 	for(j=0; j<n; j++)
483 	{
484 		if ((u_char)bp[j]<NASC)
485 			ct[bp[j]|040]++;
486 		switch (bp[j])
487 		{
488 		case '.':
489 		case ',':
490 		case ')':
491 		case '%':
492 		case ';':
493 		case ':':
494 		case '?':
495 			punct++;
496 			if ( j < n-1 &&
497 			    bp[j+1] != ' ' &&
498 			    bp[j+1] != '\n')
499 				badpun++;
500 		}
501 	}
502 	if (badpun*5 > punct)
503 		return(0);
504 	vow = ct['a'] + ct['e'] + ct['i'] + ct['o'] + ct['u'];
505 	freq = ct['e'] + ct['t'] + ct['a'] + ct['i'] + ct['o'] + ct['n'];
506 	rare = ct['v'] + ct['j'] + ct['k'] + ct['q'] + ct['x'] + ct['z'];
507 	if (2*ct[';'] > ct['e']) return(0);
508 	if ( (ct['>']+ct['<']+ct['/'])>ct['e']) return(0); /* shell file test */
509 	return (vow*5 >= n-ct[' '] && freq >= 10*rare);
510 }
511 
512 shellscript(buf, sb)
513 	char buf[];
514 	struct stat *sb;
515 {
516 	register char *tp;
517 	char *cp, *xp, *index();
518 
519 	cp = index(buf, '\n');
520 	if (cp == 0 || cp - buf > in)
521 		return (0);
522 	for (tp = buf; tp != cp && isspace(*tp); tp++)
523 		if (!isascii(*tp))
524 			return (0);
525 	for (xp = tp; tp != cp && !isspace(*tp); tp++)
526 		if (!isascii(*tp))
527 			return (0);
528 	if (tp == xp)
529 		return (0);
530 	if (sb->st_mode & S_ISUID)
531 		printf("set-uid ");
532 	if (sb->st_mode & S_ISGID)
533 		printf("set-gid ");
534 	if (strncmp(xp, "/bin/sh", tp-xp) == 0)
535 		xp = "shell";
536 	else if (strncmp(xp, "/bin/csh", tp-xp) == 0)
537 		xp = "c-shell";
538 	else
539 		*tp = '\0';
540 	printf("executable %s script\n", xp);
541 	return (1);
542 }
543 
544 shell(bp, n, tab)
545 	char *bp;
546 	int n;
547 	char *tab[];
548 {
549 
550 	i = 0;
551 	do {
552 		if (buf[i] == '#' || buf[i] == ':')
553 			while (i < n && buf[i] != '\n')
554 				i++;
555 		if (++i >= n)
556 			break;
557 		if (lookup(tab) == 1)
558 			return (1);
559 	} while (i < n);
560 	return (0);
561 }
562