[unix-history] / usr / src / usr.bin / file / file.c

static	char sccsid[] = "@(#)file.c	4.4 (Berkeley) 4.4";
/*
 * file - determine type of file
 */

#include <pagsiz.h>
#include <sys/types.h>
#include <stat.h>
#include <stdio.h>
#include <ctype.h>
#include <a.out.h>
int in;
int i  = 0;
char buf[BUFSIZ];
char *troff[] = {	/* new troff intermediate lang */
	"x","T","res","init","font","202","V0","p1",0};
char *fort[] = {
	"function","subroutine","common","dimension","block","integer",
	"real","data","double",0};
char *asc[] = {
	"chmk","mov","tst","clr","jmp",0};
char *c[] = {
	"int","char","float","double","struct","extern",0};
char *as[] = {
	"globl","byte","align","text","data","comm",0};
int	ifile;

main(argc, argv)
char **argv;
{
	FILE *fl;
	register char *p;
	char ap[128];
	extern char _sobuf[];

	if (argc>1 && argv[1][0]=='-' && argv[1][1]=='f') {
		if ((fl = fopen(argv[2], "r")) == NULL) {
			printf("Can't open %s\n", argv[2]);
			exit(2);
		}
		while ((p = fgets(ap, 128, fl)) != NULL) {
			int l = strlen(p);
			if (l>0)
				p[l-1] = '\0';
			printf("%s:	", p);
			type(p);
			if (ifile>=0)
				close(ifile);
		}
		exit(1);
	}
	while(argc > 1) {
		printf("%s:	", argv[1]);
		type(argv[1]);
		fflush(stdout);
		argc--;
		argv++;
		if (ifile >= 0)
			close(ifile);
	}
}

type(file)
char *file;
{
	int j,nl;
	char ch;
	struct stat mbuf;

	ifile = -1;
	if (lstat(file, &mbuf) < 0) {
		printf("cannot stat\n");
		return;
	}
	switch (mbuf.st_mode & S_IFMT) {

	case S_IFCHR:
		printf("character");
		goto spcl;

	case S_IFLNK:
		printf("symbolic link\n");
		return;

	case S_IFDIR:
		printf("directory\n");
		return;

	case S_IFBLK:
		printf("block");

spcl:
		printf(" special (%d/%d)\n", major(mbuf.st_rdev), minor(mbuf.st_rdev));
		return;
	}

	ifile = open(file, 0);
	if(ifile < 0) {
		printf("cannot open\n");
		return;
	}
	in = read(ifile, buf, BUFSIZ);
	if(in == 0){
		printf("empty\n");
		return;
	}
	switch(*(int *)buf) {

	case 0413:
		printf("demand paged ");

	case 0410:
		printf("pure ");
		goto exec;

	case 0411:
		printf("jfr or pdp-11 unix 411 executable\n");
		return;

	case 0407:
exec:
		printf("executable");
		if(((int *)buf)[4] != 0) {
			printf(" not stripped");
			if(oldo(buf))
				printf(" (old format symbol table)");
		}
		printf("\n");
		goto out;

	case 0177555:
		printf("very old archive\n");
		goto out;

	case 0177545:
		printf("old archive\n");
		goto out;

	case 070707:
		printf("cpio data\n");
		goto out;
	}

	if(strncmp(buf, "!<arch>\n__.SYMDEF", 17) == 0 ) {
		printf("archive random library\n");
		goto out;
	}
	if (strncmp(buf, "!<arch>\n", 8)==0) {
		printf("archive\n");
		goto out;
	}
	if (mbuf.st_size % 512 == 0) {	/* it may be a PRESS file */
		lseek(ifile, -512L, 2);	/* last block */
		if (read(ifile, buf, BUFSIZ) > 0
		 && *(short int *)buf == 12138) {
			printf("PRESS file\n");
			goto out;
		}
	}
	i = 0;
	if(ccom() == 0)goto notc;
	while(buf[i] == '#'){
		j = i;
		while(buf[i++] != '\n'){
			if(i - j > 255){
				printf("data\n"); 
				goto out;
			}
			if(i >= in)goto notc;
		}
		if(ccom() == 0)goto notc;
	}
check:
	if(lookup(c) == 1){
		while((ch = buf[i++]) != ';' && ch != '{')if(i >= in)goto notc;
		printf("c program text");
		goto outa;
	}
	nl = 0;
	while(buf[i] != '('){
		if(buf[i] <= 0)
			goto notas;
		if(buf[i] == ';'){
			i++; 
			goto check; 
		}
		if(buf[i++] == '\n')
			if(nl++ > 6)goto notc;
		if(i >= in)goto notc;
	}
	while(buf[i] != ')'){
		if(buf[i++] == '\n')
			if(nl++ > 6)goto notc;
		if(i >= in)goto notc;
	}
	while(buf[i] != '{'){
		if(buf[i++] == '\n')
			if(nl++ > 6)goto notc;
		if(i >= in)goto notc;
	}
	printf("c program text");
	goto outa;
notc:
	i = 0;
	while(buf[i] == 'c' || buf[i] == '#'){
		while(buf[i++] != '\n')if(i >= in)goto notfort;
	}
	if(lookup(fort) == 1){
		printf("fortran program text");
		goto outa;
	}
notfort:
	i=0;
	if(ascom() == 0)goto notas;
	j = i-1;
	if(buf[i] == '.'){
		i++;
		if(lookup(as) == 1){
			printf("assembler program text"); 
			goto outa;
		}
		else if(buf[j] == '\n' && isalpha(buf[j+2])){
			printf("roff, nroff, or eqn input text");
			goto outa;
		}
	}
	while(lookup(asc) == 0){
		if(ascom() == 0)goto notas;
		while(buf[i] != '\n' && buf[i++] != ':')
			if(i >= in)goto notas;
		while(buf[i] == '\n' || buf[i] == ' ' || buf[i] == '\t')if(i++ >= in)goto notas;
		j = i-1;
		if(buf[i] == '.'){
			i++;
			if(lookup(as) == 1){
				printf("assembler program text"); 
				goto outa; 
			}
			else if(buf[j] == '\n' && isalpha(buf[j+2])){
				printf("roff, nroff, or eqn input text");
				goto outa;
			}
		}
	}
	printf("assembler program text");
	goto outa;
notas:
	for(i=0; i < in; i++)if(buf[i]&0200){
		if (buf[0]=='\100' && buf[1]=='\357') {
			printf("troff (CAT) output\n");
			goto out;
		}
		printf("data\n"); 
		goto out; 
	}
	if (mbuf.st_mode&((S_IEXEC)|(S_IEXEC>>3)|(S_IEXEC>>6)))
		printf("commands text");
	else if (troffint(buf, in))
		printf("troff intermediate output text");
	else if (english(buf, in))
		printf("English text");
	else
		printf("ascii text");
outa:
	while(i < in)
		if((buf[i++]&0377) > 127){
			printf(" with garbage\n");
			goto out;
		}
	/* if next few lines in then read whole file looking for nulls ...
		while((in = read(ifile,buf,BUFSIZ)) > 0)
			for(i = 0; i < in; i++)
				if((buf[i]&0377) > 127){
					printf(" with garbage\n");
					goto out;
				}
		/*.... */
	printf("\n");
out:;
}

oldo(cp)
char *cp;
{
	struct exec ex;
	struct stat stb;

	ex = *(struct exec *)cp;
	if (fstat(ifile, &stb) < 0)
		return(0);
	if (N_STROFF(ex)+sizeof(off_t) > stb.st_size)
		return (1);
	return (0);
}


troffint(bp, n)
char *bp;
int n;
{
	int k;

	i = 0;
	for (k = 0; k < 6; k++) {
		if (lookup(troff) == 0)
			return(0);
		if (lookup(troff) == 0)
			return(0);
		while (i < n && buf[i] != '\n')
			i++;
		if (i++ >= n)
			return(0);
	}
	return(1);
}
lookup(tab)
char *tab[];
{
	char r;
	int k,j,l;
	while(buf[i] == ' ' || buf[i] == '\t' || buf[i] == '\n')i++;
	for(j=0; tab[j] != 0; j++){
		l=0;
		for(k=i; ((r=tab[j][l++]) == buf[k] && r != '\0');k++);
		if(r == '\0')
			if(buf[k] == ' ' || buf[k] == '\n' || buf[k] == '\t'
			    || buf[k] == '{' || buf[k] == '/'){
				i=k;
				return(1);
			}
	}
	return(0);
}
ccom(){
	char cc;
	while((cc = buf[i]) == ' ' || cc == '\t' || cc == '\n')if(i++ >= in)return(0);
	if(buf[i] == '/' && buf[i+1] == '*'){
		i += 2;
		while(buf[i] != '*' || buf[i+1] != '/'){
			if(buf[i] == '\\')i += 2;
			else i++;
			if(i >= in)return(0);
		}
		if((i += 2) >= in)return(0);
	}
	if(buf[i] == '\n')if(ccom() == 0)return(0);
	return(1);
}
ascom(){
	while(buf[i] == '/'){
		i++;
		while(buf[i++] != '\n')if(i >= in)return(0);
		while(buf[i] == '\n')if(i++ >= in)return(0);
	}
	return(1);
}

english (bp, n)
char *bp;
{
# define NASC 128
	int ct[NASC], j, vow, freq, rare;
	int badpun = 0, punct = 0;
	if (n<50) return(0); /* no point in statistics on squibs */
	for(j=0; j<NASC; j++)
		ct[j]=0;
	for(j=0; j<n; j++)
	{
		if (bp[j]<NASC)
			ct[bp[j]|040]++;
		switch (bp[j])
		{
		case '.': 
		case ',': 
		case ')': 
		case '%':
		case ';': 
		case ':': 
		case '?':
			punct++;
			if ( j < n-1 &&
			    bp[j+1] != ' ' &&
			    bp[j+1] != '\n')
				badpun++;
		}
	}
	if (badpun*5 > punct)
		return(0);
	vow = ct['a'] + ct['e'] + ct['i'] + ct['o'] + ct['u'];
	freq = ct['e'] + ct['t'] + ct['a'] + ct['i'] + ct['o'] + ct['n'];
	rare = ct['v'] + ct['j'] + ct['k'] + ct['q'] + ct['x'] + ct['z'];
	if (2*ct[';'] > ct['e']) return(0);
	if ( (ct['>']+ct['<']+ct['/'])>ct['e']) return(0); /* shell file test */
	return (vow*5 >= n-ct[' '] && freq >= 10*rare);
}
Commit	Line	Data
2db3d63d	1	static char sccsid[] = "@(#)file.c 4.4 (Berkeley) 4.4";
fd2fc6c6 BJ	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>
75ea79ff	11	#include <a.out.h>
fd2fc6c6 BJ	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;
2db3d63d	71	if (lstat(file, &mbuf) < 0) {
fd2fc6c6 BJ	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
2db3d63d KM	81	case S_IFLNK:
2db3d63d KM	82	printf("symbolic link\n");
fd2fc6c6 BJ	83	return;
fd2fc6c6 BJ	84
2db3d63d KM	85	case S_IFDIR:
2db3d63d KM	86	printf("directory\n");
fd2fc6c6 BJ	87	return;
	88
	89	case S_IFBLK:
	90	printf("block");
	91
	92	spcl:
	93	printf(" special (%d/%d)\n", major(mbuf.st_rdev), minor(mbuf.st_rdev));
	94	return;
	95	}
	96
	97	ifile = open(file, 0);
	98	if(ifile < 0) {
	99	printf("cannot open\n");
	100	return;
	101	}
	102	in = read(ifile, buf, BUFSIZ);
	103	if(in == 0){
	104	printf("empty\n");
	105	return;
	106	}
	107	switch((int )buf) {
	108
	109	case 0413:
	110	printf("demand paged ");
	111
	112	case 0410:
	113	printf("pure ");
	114	goto exec;
	115
	116	case 0411:
	117	printf("jfr or pdp-11 unix 411 executable\n");
	118	return;
	119
	120	case 0407:
	121	exec:
	122	printf("executable");
	123	if(((int *)buf)[4] != 0) {
	124	printf(" not stripped");
	125	if(oldo(buf))
	126	printf(" (old format symbol table)");
	127	}
	128	printf("\n");
	129	goto out;
	130
	131	case 0177555:
	132	printf("very old archive\n");
	133	goto out;
	134
	135	case 0177545:
	136	printf("old archive\n");
	137	goto out;
c083f459 BJ	138
	139	case 070707:
	140	printf("cpio data\n");
	141	goto out;
fd2fc6c6 BJ	142	}
	143
	144	if(strncmp(buf, "!<arch>\n__.SYMDEF", 17) == 0 ) {
	145	printf("archive random library\n");
	146	goto out;
	147	}
	148	if (strncmp(buf, "!<arch>\n", 8)==0) {
	149	printf("archive\n");
	150	goto out;
	151	}
47aeb93c BJ	152	if (mbuf.st_size % 512 == 0) { /* it may be a PRESS file */
	153	lseek(ifile, -512L, 2); /* last block */
	154	if (read(ifile, buf, BUFSIZ) > 0
	155	&& (short int )buf == 12138) {
	156	printf("PRESS file\n");
	157	goto out;
	158	}
	159	}
fd2fc6c6 BJ	160	i = 0;
	161	if(ccom() == 0)goto notc;
	162	while(buf[i] == '#'){
	163	j = i;
	164	while(buf[i++] != '\n'){
	165	if(i - j > 255){
	166	printf("data\n");
	167	goto out;
	168	}
	169	if(i >= in)goto notc;
	170	}
	171	if(ccom() == 0)goto notc;
	172	}
	173	check:
	174	if(lookup(c) == 1){
	175	while((ch = buf[i++]) != ';' && ch != '{')if(i >= in)goto notc;
	176	printf("c program text");
	177	goto outa;
	178	}
	179	nl = 0;
	180	while(buf[i] != '('){
	181	if(buf[i] <= 0)
	182	goto notas;
	183	if(buf[i] == ';'){
	184	i++;
	185	goto check;
	186	}
	187	if(buf[i++] == '\n')
	188	if(nl++ > 6)goto notc;
	189	if(i >= in)goto notc;
	190	}
	191	while(buf[i] != ')'){
	192	if(buf[i++] == '\n')
	193	if(nl++ > 6)goto notc;
	194	if(i >= in)goto notc;
	195	}
	196	while(buf[i] != '{'){
	197	if(buf[i++] == '\n')
	198	if(nl++ > 6)goto notc;
	199	if(i >= in)goto notc;
	200	}
	201	printf("c program text");
	202	goto outa;
	203	notc:
	204	i = 0;
	205	while(buf[i] == 'c' \|\| buf[i] == '#'){
	206	while(buf[i++] != '\n')if(i >= in)goto notfort;
	207	}
	208	if(lookup(fort) == 1){
	209	printf("fortran program text");
	210	goto outa;
	211	}
	212	notfort:
	213	i=0;
	214	if(ascom() == 0)goto notas;
	215	j = i-1;
	216	if(buf[i] == '.'){
	217	i++;
	218	if(lookup(as) == 1){
	219	printf("assembler program text");
	220	goto outa;
	221	}
	222	else if(buf[j] == '\n' && isalpha(buf[j+2])){
	223	printf("roff, nroff, or eqn input text");
224	goto outa;
225	}
226	}
227	while(lookup(asc) == 0){
228	if(ascom() == 0)goto notas;
229	while(buf[i] != '\n' && buf[i++] != ':')
230	if(i >= in)goto notas;
231	while(buf[i] == '\n' \|\| buf[i] == ' ' \|\| buf[i] == '\t')if(i++ >= in)goto notas;
232	j = i-1;
233	if(buf[i] == '.'){
234	i++;
235	if(lookup(as) == 1){
236	printf("assembler program text");
237	goto outa;
238	}
239	else if(buf[j] == '\n' && isalpha(buf[j+2])){
240	printf("roff, nroff, or eqn input text");
241	goto outa;
242	}
243	}
244	}
245	printf("assembler program text");
246	goto outa;
247	notas:
248	for(i=0; i < in; i++)if(buf[i]&0200){
249	if (buf[0]=='\100' && buf[1]=='\357') {
250	printf("troff (CAT) output\n");
251	goto out;
252	}
253	printf("data\n");
254	goto out;
255	}
256	if (mbuf.st_mode&((S_IEXEC)\|(S_IEXEC>>3)\|(S_IEXEC>>6)))
257	printf("commands text");
258	else if (troffint(buf, in))
259	printf("troff intermediate output text");
260	else if (english(buf, in))
261	printf("English text");
262	else
263	printf("ascii text");
264	outa:
265	while(i < in)
266	if((buf[i++]&0377) > 127){
267	printf(" with garbage\n");
268	goto out;
269	}
270	/* if next few lines in then read whole file looking for nulls ...
271	while((in = read(ifile,buf,BUFSIZ)) > 0)
272	for(i = 0; i < in; i++)
273	if((buf[i]&0377) > 127){
274	printf(" with garbage\n");
275	goto out;
276	}
277	/.... /
278	printf("\n");
279	out:;
280	}
281
282	oldo(cp)
283	char *cp;
284	{
285	struct exec ex;
286	struct stat stb;
287
288	ex = (struct exec )cp;
289	if (fstat(ifile, &stb) < 0)
290	return(0);
291	if (N_STROFF(ex)+sizeof(off_t) > stb.st_size)
292	return (1);
293	return (0);
294	}
295
296
297
298	troffint(bp, n)
299	char *bp;
300	int n;
301	{
302	int k;
303
304	i = 0;
305	for (k = 0; k < 6; k++) {
306	if (lookup(troff) == 0)
307	return(0);
308	if (lookup(troff) == 0)
309	return(0);
310	while (i < n && buf[i] != '\n')
311	i++;
312	if (i++ >= n)
313	return(0);
314	}
315	return(1);
316	}
317	lookup(tab)
318	char *tab[];
319	{
320	char r;
321	int k,j,l;
322	while(buf[i] == ' ' \|\| buf[i] == '\t' \|\| buf[i] == '\n')i++;
323	for(j=0; tab[j] != 0; j++){
324	l=0;
325	for(k=i; ((r=tab[j][l++]) == buf[k] && r != '\0');k++);
326	if(r == '\0')
327	if(buf[k] == ' ' \|\| buf[k] == '\n' \|\| buf[k] == '\t'
328	\|\| buf[k] == '{' \|\| buf[k] == '/'){
329	i=k;
330	return(1);
331	}
332	}
333	return(0);
334	}
335	ccom(){
336	char cc;
337	while((cc = buf[i]) == ' ' \|\| cc == '\t' \|\| cc == '\n')if(i++ >= in)return(0);
338	if(buf[i] == '/' && buf[i+1] == '*'){
339	i += 2;
340	while(buf[i] != '*' \|\| buf[i+1] != '/'){
341	if(buf[i] == '\\')i += 2;
342	else i++;
343	if(i >= in)return(0);
344	}
345	if((i += 2) >= in)return(0);
346	}
347	if(buf[i] == '\n')if(ccom() == 0)return(0);
348	return(1);
349	}
350	ascom(){
351	while(buf[i] == '/'){
352	i++;
353	while(buf[i++] != '\n')if(i >= in)return(0);
354	while(buf[i] == '\n')if(i++ >= in)return(0);
355	}
356	return(1);
357	}
358
359	english (bp, n)
360	char *bp;
361	{
362	# define NASC 128
363	int ct[NASC], j, vow, freq, rare;
364	int badpun = 0, punct = 0;
365	if (n<50) return(0); /* no point in statistics on squibs */
366	for(j=0; j<NASC; j++)
367	ct[j]=0;
368	for(j=0; j<n; j++)
369	{
370	if (bp[j]<NASC)
371	ct[bp[j]\|040]++;
372	switch (bp[j])
373	{
374	case '.':
375	case ',':
376	case ')':
377	case '%':
378	case ';':
379	case ':':
380	case '?':
381	punct++;
382	if ( j < n-1 &&
383	bp[j+1] != ' ' &&
384	bp[j+1] != '\n')
385	badpun++;
386	}
387	}
388	if (badpun*5 > punct)
389	return(0);
390	vow = ct['a'] + ct['e'] + ct['i'] + ct['o'] + ct['u'];
391	freq = ct['e'] + ct['t'] + ct['a'] + ct['i'] + ct['o'] + ct['n'];
392	rare = ct['v'] + ct['j'] + ct['k'] + ct['q'] + ct['x'] + ct['z'];
393	if (2*ct[';'] > ct['e']) return(0);
394	if ( (ct['>']+ct['<']+ct['/'])>ct['e']) return(0); /* shell file test */
395	return (vow5 >= n-ct[' '] && freq >= 10rare);
396	}