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

#ifndef lint
static	char sccsid[] = "@(#)file.c	4.6 (Berkeley) 4.6";
#endif
/*
 * file - determine type of file
 */

#include <sys/types.h>
#include <stat.h>
#include <stdio.h>
#include <ctype.h>
#include <a.out.h>
int	errno;
char	*sys_errlist[];
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) {
			perror(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("%s\n", sys_errlist[errno]);
		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
34aa0635 SL	1	#ifndef lint
	2	static char sccsid[] = "@(#)file.c 4.6 (Berkeley) 4.6";
	3	#endif
fd2fc6c6 BJ	4	/*
	5	* file - determine type of file
	6	*/
	7
fd2fc6c6 BJ	8	#include <sys/types.h>
	9	#include <stat.h>
	10	#include <stdio.h>
	11	#include <ctype.h>
75ea79ff	12	#include <a.out.h>
136b085b KM	13	int errno;
136b085b KM	14	char *sys_errlist[];
fd2fc6c6 BJ	15	int in;
	16	int i = 0;
	17	char buf[BUFSIZ];
	18	char troff[] = { / new troff intermediate lang */
	19	"x","T","res","init","font","202","V0","p1",0};
	20	char *fort[] = {
	21	"function","subroutine","common","dimension","block","integer",
	22	"real","data","double",0};
	23	char *asc[] = {
	24	"chmk","mov","tst","clr","jmp",0};
	25	char *c[] = {
	26	"int","char","float","double","struct","extern",0};
	27	char *as[] = {
	28	"globl","byte","align","text","data","comm",0};
	29	int ifile;
	30
	31	main(argc, argv)
	32	char **argv;
	33	{
	34	FILE *fl;
	35	register char *p;
	36	char ap[128];
	37	extern char _sobuf[];
	38
	39	if (argc>1 && argv[1][0]=='-' && argv[1][1]=='f') {
	40	if ((fl = fopen(argv[2], "r")) == NULL) {
136b085b	41	perror(argv[2]);
fd2fc6c6 BJ	42	exit(2);
	43	}
	44	while ((p = fgets(ap, 128, fl)) != NULL) {
	45	int l = strlen(p);
	46	if (l>0)
	47	p[l-1] = '\0';
	48	printf("%s: ", p);
	49	type(p);
	50	if (ifile>=0)
	51	close(ifile);
	52	}
	53	exit(1);
	54	}
	55	while(argc > 1) {
	56	printf("%s: ", argv[1]);
	57	type(argv[1]);
	58	fflush(stdout);
	59	argc--;
	60	argv++;
	61	if (ifile >= 0)
	62	close(ifile);
	63	}
	64	}
	65
	66	type(file)
	67	char *file;
	68	{
	69	int j,nl;
	70	char ch;
	71	struct stat mbuf;
	72
	73	ifile = -1;
2db3d63d	74	if (lstat(file, &mbuf) < 0) {
136b085b	75	printf("%s\n", sys_errlist[errno]);
fd2fc6c6 BJ	76	return;
	77	}
	78	switch (mbuf.st_mode & S_IFMT) {
	79
	80	case S_IFCHR:
	81	printf("character");
	82	goto spcl;
	83
2db3d63d KM	84	case S_IFLNK:
2db3d63d KM	85	printf("symbolic link\n");
fd2fc6c6 BJ	86	return;
fd2fc6c6 BJ	87
2db3d63d KM	88	case S_IFDIR:
2db3d63d KM	89	printf("directory\n");
fd2fc6c6 BJ	90	return;
	91
	92	case S_IFBLK:
	93	printf("block");
	94
	95	spcl:
	96	printf(" special (%d/%d)\n", major(mbuf.st_rdev), minor(mbuf.st_rdev));
	97	return;
	98	}
	99
	100	ifile = open(file, 0);
	101	if(ifile < 0) {
	102	printf("cannot open\n");
	103	return;
	104	}
	105	in = read(ifile, buf, BUFSIZ);
	106	if(in == 0){
	107	printf("empty\n");
	108	return;
	109	}
	110	switch((int )buf) {
	111
	112	case 0413:
	113	printf("demand paged ");
	114
	115	case 0410:
	116	printf("pure ");
	117	goto exec;
	118
	119	case 0411:
	120	printf("jfr or pdp-11 unix 411 executable\n");
	121	return;
	122
	123	case 0407:
	124	exec:
	125	printf("executable");
	126	if(((int *)buf)[4] != 0) {
	127	printf(" not stripped");
	128	if(oldo(buf))
	129	printf(" (old format symbol table)");
	130	}
	131	printf("\n");
	132	goto out;
	133
	134	case 0177555:
	135	printf("very old archive\n");
	136	goto out;
	137
	138	case 0177545:
	139	printf("old archive\n");
	140	goto out;
c083f459 BJ	141
	142	case 070707:
	143	printf("cpio data\n");
	144	goto out;
fd2fc6c6 BJ	145	}
	146
	147	if(strncmp(buf, "!<arch>\n__.SYMDEF", 17) == 0 ) {
	148	printf("archive random library\n");
	149	goto out;
	150	}
	151	if (strncmp(buf, "!<arch>\n", 8)==0) {
	152	printf("archive\n");
	153	goto out;
	154	}
47aeb93c BJ	155	if (mbuf.st_size % 512 == 0) { /* it may be a PRESS file */
	156	lseek(ifile, -512L, 2); /* last block */
	157	if (read(ifile, buf, BUFSIZ) > 0
	158	&& (short int )buf == 12138) {
	159	printf("PRESS file\n");
	160	goto out;
	161	}
	162	}
fd2fc6c6 BJ	163	i = 0;
	164	if(ccom() == 0)goto notc;
	165	while(buf[i] == '#'){
	166	j = i;
	167	while(buf[i++] != '\n'){
	168	if(i - j > 255){
	169	printf("data\n");
	170	goto out;
	171	}
	172	if(i >= in)goto notc;
	173	}
	174	if(ccom() == 0)goto notc;
	175	}
	176	check:
	177	if(lookup(c) == 1){
	178	while((ch = buf[i++]) != ';' && ch != '{')if(i >= in)goto notc;
	179	printf("c program text");
	180	goto outa;
	181	}
	182	nl = 0;
	183	while(buf[i] != '('){
	184	if(buf[i] <= 0)
	185	goto notas;
	186	if(buf[i] == ';'){
	187	i++;
	188	goto check;
	189	}
	190	if(buf[i++] == '\n')
	191	if(nl++ > 6)goto notc;
	192	if(i >= in)goto notc;
	193	}
	194	while(buf[i] != ')'){
	195	if(buf[i++] == '\n')
	196	if(nl++ > 6)goto notc;
	197	if(i >= in)goto notc;
	198	}
	199	while(buf[i] != '{'){
	200	if(buf[i++] == '\n')
	201	if(nl++ > 6)goto notc;
	202	if(i >= in)goto notc;
	203	}
	204	printf("c program text");
	205	goto outa;
	206	notc:
	207	i = 0;
	208	while(buf[i] == 'c' \|\| buf[i] == '#'){
	209	while(buf[i++] != '\n')if(i >= in)goto notfort;
	210	}
	211	if(lookup(fort) == 1){
	212	printf("fortran program text");
	213	goto outa;
	214	}
	215	notfort:
	216	i=0;
	217	if(ascom() == 0)goto notas;
	218	j = i-1;
	219	if(buf[i] == '.'){
	220	i++;
	221	if(lookup(as) == 1){
	222	printf("assembler program text");
	223	goto outa;
	224	}
	225	else if(buf[j] == '\n' && isalpha(buf[j+2])){
	226	printf("roff, nroff, or eqn input text");
227	goto outa;
228	}
229	}
230	while(lookup(asc) == 0){
231	if(ascom() == 0)goto notas;
232	while(buf[i] != '\n' && buf[i++] != ':')
233	if(i >= in)goto notas;
234	while(buf[i] == '\n' \|\| buf[i] == ' ' \|\| buf[i] == '\t')if(i++ >= in)goto notas;
235	j = i-1;
236	if(buf[i] == '.'){
237	i++;
238	if(lookup(as) == 1){
239	printf("assembler program text");
240	goto outa;
241	}
242	else if(buf[j] == '\n' && isalpha(buf[j+2])){
243	printf("roff, nroff, or eqn input text");
244	goto outa;
245	}
246	}
247	}
248	printf("assembler program text");
249	goto outa;
250	notas:
251	for(i=0; i < in; i++)if(buf[i]&0200){
252	if (buf[0]=='\100' && buf[1]=='\357') {
253	printf("troff (CAT) output\n");
254	goto out;
255	}
256	printf("data\n");
257	goto out;
258	}
259	if (mbuf.st_mode&((S_IEXEC)\|(S_IEXEC>>3)\|(S_IEXEC>>6)))
260	printf("commands text");
261	else if (troffint(buf, in))
262	printf("troff intermediate output text");
263	else if (english(buf, in))
264	printf("English text");
265	else
266	printf("ascii text");
267	outa:
268	while(i < in)
269	if((buf[i++]&0377) > 127){
270	printf(" with garbage\n");
271	goto out;
272	}
273	/* if next few lines in then read whole file looking for nulls ...
274	while((in = read(ifile,buf,BUFSIZ)) > 0)
275	for(i = 0; i < in; i++)
276	if((buf[i]&0377) > 127){
277	printf(" with garbage\n");
278	goto out;
279	}
280	/.... /
281	printf("\n");
282	out:;
283	}
284
285	oldo(cp)
286	char *cp;
287	{
288	struct exec ex;
289	struct stat stb;
290
291	ex = (struct exec )cp;
292	if (fstat(ifile, &stb) < 0)
293	return(0);
294	if (N_STROFF(ex)+sizeof(off_t) > stb.st_size)
295	return (1);
296	return (0);
297	}
298
299
300
301	troffint(bp, n)
302	char *bp;
303	int n;
304	{
305	int k;
306
307	i = 0;
308	for (k = 0; k < 6; k++) {
309	if (lookup(troff) == 0)
310	return(0);
311	if (lookup(troff) == 0)
312	return(0);
313	while (i < n && buf[i] != '\n')
314	i++;
315	if (i++ >= n)
316	return(0);
317	}
318	return(1);
319	}
320	lookup(tab)
321	char *tab[];
322	{
323	char r;
324	int k,j,l;
325	while(buf[i] == ' ' \|\| buf[i] == '\t' \|\| buf[i] == '\n')i++;
326	for(j=0; tab[j] != 0; j++){
327	l=0;
328	for(k=i; ((r=tab[j][l++]) == buf[k] && r != '\0');k++);
329	if(r == '\0')
330	if(buf[k] == ' ' \|\| buf[k] == '\n' \|\| buf[k] == '\t'
331	\|\| buf[k] == '{' \|\| buf[k] == '/'){
332	i=k;
333	return(1);
334	}
335	}
336	return(0);
337	}
338	ccom(){
339	char cc;
340	while((cc = buf[i]) == ' ' \|\| cc == '\t' \|\| cc == '\n')if(i++ >= in)return(0);
341	if(buf[i] == '/' && buf[i+1] == '*'){
342	i += 2;
343	while(buf[i] != '*' \|\| buf[i+1] != '/'){
344	if(buf[i] == '\\')i += 2;
345	else i++;
346	if(i >= in)return(0);
347	}
348	if((i += 2) >= in)return(0);
349	}
350	if(buf[i] == '\n')if(ccom() == 0)return(0);
351	return(1);
352	}
353	ascom(){
354	while(buf[i] == '/'){
355	i++;
356	while(buf[i++] != '\n')if(i >= in)return(0);
357	while(buf[i] == '\n')if(i++ >= in)return(0);
358	}
359	return(1);
360	}
361
362	english (bp, n)
363	char *bp;
364	{
365	# define NASC 128
366	int ct[NASC], j, vow, freq, rare;
367	int badpun = 0, punct = 0;
368	if (n<50) return(0); /* no point in statistics on squibs */
369	for(j=0; j<NASC; j++)
370	ct[j]=0;
371	for(j=0; j<n; j++)
372	{
373	if (bp[j]<NASC)
374	ct[bp[j]\|040]++;
375	switch (bp[j])
376	{
377	case '.':
378	case ',':
379	case ')':
380	case '%':
381	case ';':
382	case ':':
383	case '?':
384	punct++;
385	if ( j < n-1 &&
386	bp[j+1] != ' ' &&
387	bp[j+1] != '\n')
388	badpun++;
389	}
390	}
391	if (badpun*5 > punct)
392	return(0);
393	vow = ct['a'] + ct['e'] + ct['i'] + ct['o'] + ct['u'];
394	freq = ct['e'] + ct['t'] + ct['a'] + ct['i'] + ct['o'] + ct['n'];
395	rare = ct['v'] + ct['j'] + ct['k'] + ct['q'] + ct['x'] + ct['z'];
396	if (2*ct[';'] > ct['e']) return(0);
397	if ( (ct['>']+ct['<']+ct['/'])>ct['e']) return(0); /* shell file test */
398	return (vow5 >= n-ct[' '] && freq >= 10rare);
399	}