xref: /original-bsd/usr.bin/file/file.c (revision 552e81d8)
1 static	char sccsid[] = "@(#)file.c 4.1 10/01/80";
2 /*
3  * file - determine type of file
4  */
5 
6 #include <pagsiz.h>
7 #include <sys/types.h>
8 #include <stat.h>
9 #include <stdio.h>
10 #include <ctype.h>
11 #include <a.out.h>
12 int in;
13 int i  = 0;
14 char buf[BUFSIZ];
15 char *troff[] = {	/* new troff intermediate lang */
16 	"x","T","res","init","font","202","V0","p1",0};
17 char *fort[] = {
18 	"function","subroutine","common","dimension","block","integer",
19 	"real","data","double",0};
20 char *asc[] = {
21 	"chmk","mov","tst","clr","jmp",0};
22 char *c[] = {
23 	"int","char","float","double","struct","extern",0};
24 char *as[] = {
25 	"globl","byte","align","text","data","comm",0};
26 int	ifile;
27 
28 main(argc, argv)
29 char **argv;
30 {
31 	FILE *fl;
32 	register char *p;
33 	char ap[128];
34 	extern char _sobuf[];
35 
36 	if (argc>1 && argv[1][0]=='-' && argv[1][1]=='f') {
37 		if ((fl = fopen(argv[2], "r")) == NULL) {
38 			printf("Can't open %s\n", argv[2]);
39 			exit(2);
40 		}
41 		while ((p = fgets(ap, 128, fl)) != NULL) {
42 			int l = strlen(p);
43 			if (l>0)
44 				p[l-1] = '\0';
45 			printf("%s:	", p);
46 			type(p);
47 			if (ifile>=0)
48 				close(ifile);
49 		}
50 		exit(1);
51 	}
52 	while(argc > 1) {
53 		printf("%s:	", argv[1]);
54 		type(argv[1]);
55 		fflush(stdout);
56 		argc--;
57 		argv++;
58 		if (ifile >= 0)
59 			close(ifile);
60 	}
61 }
62 
63 type(file)
64 char *file;
65 {
66 	int j,nl;
67 	char ch;
68 	struct stat mbuf;
69 
70 	ifile = -1;
71 	if(stat(file, &mbuf) < 0) {
72 		printf("cannot stat\n");
73 		return;
74 	}
75 	switch (mbuf.st_mode & S_IFMT) {
76 
77 	case S_IFCHR:
78 		printf("character");
79 		goto spcl;
80 
81 	case S_IFDIR:
82 		printf("directory\n");
83 		return;
84 
85 	case S_IFMPC:
86 		printf("char multiplexor\n");
87 		return;
88 
89 	case S_IFMPB:
90 		printf("block multiplexor\n");
91 		return;
92 
93 	case S_IFBLK:
94 		printf("block");
95 
96 spcl:
97 		printf(" special (%d/%d)\n", major(mbuf.st_rdev), minor(mbuf.st_rdev));
98 		return;
99 	}
100 
101 	ifile = open(file, 0);
102 	if(ifile < 0) {
103 		printf("cannot open\n");
104 		return;
105 	}
106 	in = read(ifile, buf, BUFSIZ);
107 	if(in == 0){
108 		printf("empty\n");
109 		return;
110 	}
111 	switch(*(int *)buf) {
112 
113 	case 0413:
114 		printf("demand paged ");
115 
116 	case 0410:
117 		printf("pure ");
118 		goto exec;
119 
120 	case 0411:
121 		printf("jfr or pdp-11 unix 411 executable\n");
122 		return;
123 
124 	case 0407:
125 exec:
126 		printf("executable");
127 		if(((int *)buf)[4] != 0) {
128 			printf(" not stripped");
129 			if(oldo(buf))
130 				printf(" (old format symbol table)");
131 		}
132 		printf("\n");
133 		goto out;
134 
135 	case 0177555:
136 		printf("very old archive\n");
137 		goto out;
138 
139 	case 0177545:
140 		printf("old archive\n");
141 		goto out;
142 
143 	case 070707:
144 		printf("cpio data\n");
145 		goto out;
146 	}
147 
148 	if(strncmp(buf, "!<arch>\n__.SYMDEF", 17) == 0 ) {
149 		printf("archive random library\n");
150 		goto out;
151 	}
152 	if (strncmp(buf, "!<arch>\n", 8)==0) {
153 		printf("archive\n");
154 		goto out;
155 	}
156 	i = 0;
157 	if(ccom() == 0)goto notc;
158 	while(buf[i] == '#'){
159 		j = i;
160 		while(buf[i++] != '\n'){
161 			if(i - j > 255){
162 				printf("data\n");
163 				goto out;
164 			}
165 			if(i >= in)goto notc;
166 		}
167 		if(ccom() == 0)goto notc;
168 	}
169 check:
170 	if(lookup(c) == 1){
171 		while((ch = buf[i++]) != ';' && ch != '{')if(i >= in)goto notc;
172 		printf("c program text");
173 		goto outa;
174 	}
175 	nl = 0;
176 	while(buf[i] != '('){
177 		if(buf[i] <= 0)
178 			goto notas;
179 		if(buf[i] == ';'){
180 			i++;
181 			goto check;
182 		}
183 		if(buf[i++] == '\n')
184 			if(nl++ > 6)goto notc;
185 		if(i >= in)goto notc;
186 	}
187 	while(buf[i] != ')'){
188 		if(buf[i++] == '\n')
189 			if(nl++ > 6)goto notc;
190 		if(i >= in)goto notc;
191 	}
192 	while(buf[i] != '{'){
193 		if(buf[i++] == '\n')
194 			if(nl++ > 6)goto notc;
195 		if(i >= in)goto notc;
196 	}
197 	printf("c program text");
198 	goto outa;
199 notc:
200 	i = 0;
201 	while(buf[i] == 'c' || buf[i] == '#'){
202 		while(buf[i++] != '\n')if(i >= in)goto notfort;
203 	}
204 	if(lookup(fort) == 1){
205 		printf("fortran program text");
206 		goto outa;
207 	}
208 notfort:
209 	i=0;
210 	if(ascom() == 0)goto notas;
211 	j = i-1;
212 	if(buf[i] == '.'){
213 		i++;
214 		if(lookup(as) == 1){
215 			printf("assembler program text");
216 			goto outa;
217 		}
218 		else if(buf[j] == '\n' && isalpha(buf[j+2])){
219 			printf("roff, nroff, or eqn input text");
220 			goto outa;
221 		}
222 	}
223 	while(lookup(asc) == 0){
224 		if(ascom() == 0)goto notas;
225 		while(buf[i] != '\n' && buf[i++] != ':')
226 			if(i >= in)goto notas;
227 		while(buf[i] == '\n' || buf[i] == ' ' || buf[i] == '\t')if(i++ >= in)goto notas;
228 		j = i-1;
229 		if(buf[i] == '.'){
230 			i++;
231 			if(lookup(as) == 1){
232 				printf("assembler program text");
233 				goto outa;
234 			}
235 			else if(buf[j] == '\n' && isalpha(buf[j+2])){
236 				printf("roff, nroff, or eqn input text");
237 				goto outa;
238 			}
239 		}
240 	}
241 	printf("assembler program text");
242 	goto outa;
243 notas:
244 	for(i=0; i < in; i++)if(buf[i]&0200){
245 		if (buf[0]=='\100' && buf[1]=='\357') {
246 			printf("troff (CAT) output\n");
247 			goto out;
248 		}
249 		printf("data\n");
250 		goto out;
251 	}
252 	if (mbuf.st_mode&((S_IEXEC)|(S_IEXEC>>3)|(S_IEXEC>>6)))
253 		printf("commands text");
254 	else if (troffint(buf, in))
255 		printf("troff intermediate output text");
256 	else if (english(buf, in))
257 		printf("English text");
258 	else
259 		printf("ascii text");
260 outa:
261 	while(i < in)
262 		if((buf[i++]&0377) > 127){
263 			printf(" with garbage\n");
264 			goto out;
265 		}
266 	/* if next few lines in then read whole file looking for nulls ...
267 		while((in = read(ifile,buf,BUFSIZ)) > 0)
268 			for(i = 0; i < in; i++)
269 				if((buf[i]&0377) > 127){
270 					printf(" with garbage\n");
271 					goto out;
272 				}
273 		/*.... */
274 	printf("\n");
275 out:;
276 }
277 
278 oldo(cp)
279 char *cp;
280 {
281 	struct exec ex;
282 	struct stat stb;
283 
284 	ex = *(struct exec *)cp;
285 	if (fstat(ifile, &stb) < 0)
286 		return(0);
287 	if (N_STROFF(ex)+sizeof(off_t) > stb.st_size)
288 		return (1);
289 	return (0);
290 }
291 
292 
293 
294 troffint(bp, n)
295 char *bp;
296 int n;
297 {
298 	int k;
299 
300 	i = 0;
301 	for (k = 0; k < 6; k++) {
302 		if (lookup(troff) == 0)
303 			return(0);
304 		if (lookup(troff) == 0)
305 			return(0);
306 		while (i < n && buf[i] != '\n')
307 			i++;
308 		if (i++ >= n)
309 			return(0);
310 	}
311 	return(1);
312 }
313 lookup(tab)
314 char *tab[];
315 {
316 	char r;
317 	int k,j,l;
318 	while(buf[i] == ' ' || buf[i] == '\t' || buf[i] == '\n')i++;
319 	for(j=0; tab[j] != 0; j++){
320 		l=0;
321 		for(k=i; ((r=tab[j][l++]) == buf[k] && r != '\0');k++);
322 		if(r == '\0')
323 			if(buf[k] == ' ' || buf[k] == '\n' || buf[k] == '\t'
324 			    || buf[k] == '{' || buf[k] == '/'){
325 				i=k;
326 				return(1);
327 			}
328 	}
329 	return(0);
330 }
331 ccom(){
332 	char cc;
333 	while((cc = buf[i]) == ' ' || cc == '\t' || cc == '\n')if(i++ >= in)return(0);
334 	if(buf[i] == '/' && buf[i+1] == '*'){
335 		i += 2;
336 		while(buf[i] != '*' || buf[i+1] != '/'){
337 			if(buf[i] == '\\')i += 2;
338 			else i++;
339 			if(i >= in)return(0);
340 		}
341 		if((i += 2) >= in)return(0);
342 	}
343 	if(buf[i] == '\n')if(ccom() == 0)return(0);
344 	return(1);
345 }
346 ascom(){
347 	while(buf[i] == '/'){
348 		i++;
349 		while(buf[i++] != '\n')if(i >= in)return(0);
350 		while(buf[i] == '\n')if(i++ >= in)return(0);
351 	}
352 	return(1);
353 }
354 
355 english (bp, n)
356 char *bp;
357 {
358 # define NASC 128
359 	int ct[NASC], j, vow, freq, rare;
360 	int badpun = 0, punct = 0;
361 	if (n<50) return(0); /* no point in statistics on squibs */
362 	for(j=0; j<NASC; j++)
363 		ct[j]=0;
364 	for(j=0; j<n; j++)
365 	{
366 		if (bp[j]<NASC)
367 			ct[bp[j]|040]++;
368 		switch (bp[j])
369 		{
370 		case '.':
371 		case ',':
372 		case ')':
373 		case '%':
374 		case ';':
375 		case ':':
376 		case '?':
377 			punct++;
378 			if ( j < n-1 &&
379 			    bp[j+1] != ' ' &&
380 			    bp[j+1] != '\n')
381 				badpun++;
382 		}
383 	}
384 	if (badpun*5 > punct)
385 		return(0);
386 	vow = ct['a'] + ct['e'] + ct['i'] + ct['o'] + ct['u'];
387 	freq = ct['e'] + ct['t'] + ct['a'] + ct['i'] + ct['o'] + ct['n'];
388 	rare = ct['v'] + ct['j'] + ct['k'] + ct['q'] + ct['x'] + ct['z'];
389 	if (2*ct[';'] > ct['e']) return(0);
390 	if ( (ct['>']+ct['<']+ct['/'])>ct['e']) return(0); /* shell file test */
391 	return (vow*5 >= n-ct[' '] && freq >= 10*rare);
392 }
393