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

/*
 * Copyright (c) 1980 The Regents of the University of California.
 * All rights reserved.
 *
 * %sccs.include.redist.c%
 */

#ifndef lint
char copyright[] =
"@(#) Copyright (c) 1980 The Regents of the University of California.\n\
 All rights reserved.\n";
#endif /* not lint */

#ifndef lint
static char sccsid[] = "@(#)xstr.c	5.7 (Berkeley) %G%";
#endif /* not lint */

#include <sys/types.h>
#include <signal.h>
#include <errno.h>
#include <unistd.h>
#include <stdio.h>
#include <ctype.h>
#include <string.h>
#include "pathnames.h"

/*
 * xstr - extract and hash strings in a C program
 *
 * Bill Joy UCB
 * November, 1978
 */

#define	ignore(a)	((void) a)

off_t	tellpt;
off_t	hashit();
void	onintr();
char	*savestr();
off_t	yankstr();

off_t	mesgpt;
char	*strings =	"strings";

int	cflg;
int	vflg;
int	readstd;

main(argc, argv)
	int argc;
	char *argv[];
{

	argc--, argv++;
	while (argc > 0 && argv[0][0] == '-') {
		register char *cp = &(*argv++)[1];

		argc--;
		if (*cp == 0) {
			readstd++;
			continue;
		}
		do switch (*cp++) {

		case 'c':
			cflg++;
			continue;

		case 'v':
			vflg++;
			continue;

		default:
			fprintf(stderr, "usage: xstr [ -v ] [ -c ] [ - ] [ name ... ]\n");
		} while (*cp);
	}
	if (signal(SIGINT, SIG_IGN) == SIG_DFL)
		signal(SIGINT, onintr);
	if (cflg || argc == 0 && !readstd)
		inithash();
	else
		strings = mktemp(strdup(_PATH_TMP));
	while (readstd || argc > 0) {
		if (freopen("x.c", "w", stdout) == NULL)
			perror("x.c"), exit(1);
		if (!readstd && freopen(argv[0], "r", stdin) == NULL)
			perror(argv[0]), exit(2);
		process("x.c");
		if (readstd == 0)
			argc--, argv++;
		else
			readstd = 0;
	};
	flushsh();
	if (cflg == 0)
		xsdotc();
	if (strings[0] == '/')
		ignore(unlink(strings));
	exit(0);
}

char linebuf[BUFSIZ];

process(name)
	char *name;
{
	char *cp;
	register int c;
	register int incomm = 0;
	int ret;

	printf("extern char\txstr[];\n");
	for (;;) {
		if (fgets(linebuf, sizeof linebuf, stdin) == NULL) {
			if (ferror(stdin)) {
				perror(name);
				exit(3);
			}
			break;
		}
		if (linebuf[0] == '#') {
			if (linebuf[1] == ' ' && isdigit(linebuf[2]))
				printf("#line%s", &linebuf[1]);
			else
				printf("%s", linebuf);
			continue;
		}
		for (cp = linebuf; c = *cp++;) switch (c) {
			
		case '"':
			if (incomm)
				goto def;
			if ((ret = (int) yankstr(&cp)) == -1)
				goto out;
			printf("(&xstr[%d])", ret);
			break;

		case '\'':
			if (incomm)
				goto def;
			putchar(c);
			if (*cp)
				putchar(*cp++);
			break;

		case '/':
			if (incomm || *cp != '*')
				goto def;
			incomm = 1;
			cp++;
			printf("/*");
			continue;

		case '*':
			if (incomm && *cp == '/') {
				incomm = 0;
				cp++;
				printf("*/");
				continue;
			}
			goto def;
		
def:
		default:
			putchar(c);
			break;
		}
	}
out:
	if (ferror(stdout))
		perror("x.c"), onintr();
}

off_t
yankstr(cpp)
	register char **cpp;
{
	register char *cp = *cpp;
	register int c, ch;
	char dbuf[BUFSIZ];
	register char *dp = dbuf;
	register char *tp;

	while (c = *cp++) {
		switch (c) {

		case '"':
			cp++;
			goto out;

		case '\\':
			c = *cp++;
			if (c == 0)
				break;
			if (c == '\n') {
				if (fgets(linebuf, sizeof linebuf, stdin) 
				    == NULL) {
					if (ferror(stdin)) {
						perror("x.c");
						exit(3);
					}
					return(-1);
				}
				cp = linebuf;
				continue;
			}
			for (tp = "b\bt\tr\rn\nf\f\\\\\"\""; ch = *tp++; tp++)
				if (c == ch) {
					c = *tp;
					goto gotc;
				}
			if (!octdigit(c)) {
				*dp++ = '\\';
				break;
			}
			c -= '0';
			if (!octdigit(*cp))
				break;
			c <<= 3, c += *cp++ - '0';
			if (!octdigit(*cp))
				break;
			c <<= 3, c += *cp++ - '0';
			break;
		}
gotc:
		*dp++ = c;
	}
out:
	*cpp = --cp;
	*dp = 0;
	return (hashit(dbuf, 1));
}

octdigit(c)
	char c;
{

	return (isdigit(c) && c != '8' && c != '9');
}

inithash()
{
	char buf[BUFSIZ];
	register FILE *mesgread = fopen(strings, "r");

	if (mesgread == NULL)
		return;
	for (;;) {
		mesgpt = tellpt;
		if (fgetNUL(buf, sizeof buf, mesgread) == NULL)
			break;
		ignore(hashit(buf, 0));
	}
	ignore(fclose(mesgread));
}

fgetNUL(obuf, rmdr, file)
	char *obuf;
	register int rmdr;
	FILE *file;
{
	register c;
	register char *buf = obuf;

	while (--rmdr > 0 && (c = xgetc(file)) != 0 && c != EOF)
		*buf++ = c;
	*buf++ = 0;
	return ((feof(file) || ferror(file)) ? NULL : 1);
}

xgetc(file)
	FILE *file;
{

	tellpt++;
	return (getc(file));
}

#define	BUCKETS	128

struct	hash {
	off_t	hpt;
	char	*hstr;
	struct	hash *hnext;
	short	hnew;
} bucket[BUCKETS];

off_t
hashit(str, new)
	char *str;
	int new;
{
	int i;
	register struct hash *hp, *hp0;

	hp = hp0 = &bucket[lastchr(str) & 0177];
	while (hp->hnext) {
		hp = hp->hnext;
		i = istail(str, hp->hstr);
		if (i >= 0)
			return (hp->hpt + i);
	}
	if ((hp = (struct hash *) calloc(1, sizeof (*hp))) == NULL) {
		perror("xstr");
		exit(8);
	}
	hp->hpt = mesgpt;
	if (!(hp->hstr = strdup(str))) {
		(void)fprintf(stderr, "xstr: %s\n", strerror(errno));
		exit(1);
	}
	mesgpt += strlen(hp->hstr) + 1;
	hp->hnext = hp0->hnext;
	hp->hnew = new;
	hp0->hnext = hp;
	return (hp->hpt);
}

flushsh()
{
	register int i;
	register struct hash *hp;
	register FILE *mesgwrit;
	register int old = 0, new = 0;

	for (i = 0; i < BUCKETS; i++)
		for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext)
			if (hp->hnew)
				new++;
			else
				old++;
	if (new == 0 && old != 0)
		return;
	mesgwrit = fopen(strings, old ? "r+" : "w");
	if (mesgwrit == NULL)
		perror(strings), exit(4);
	for (i = 0; i < BUCKETS; i++)
		for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext) {
			found(hp->hnew, hp->hpt, hp->hstr);
			if (hp->hnew) {
				fseek(mesgwrit, hp->hpt, 0);
				ignore(fwrite(hp->hstr, strlen(hp->hstr) + 1, 1, mesgwrit));
				if (ferror(mesgwrit))
					perror(strings), exit(4);
			}
		}
	if (fclose(mesgwrit) == EOF)
		perror(strings), exit(4);
}

found(new, off, str)
	int new;
	off_t off;
	char *str;
{
	if (vflg == 0)
		return;
	if (!new)
		fprintf(stderr, "found at %d:", (int) off);
	else
		fprintf(stderr, "new at %d:", (int) off);
	prstr(str);
	fprintf(stderr, "\n");
}

prstr(cp)
	register char *cp;
{
	register int c;

	while (c = (*cp++ & 0377))
		if (c < ' ')
			fprintf(stderr, "^%c", c + '`');
		else if (c == 0177)
			fprintf(stderr, "^?");
		else if (c > 0200)
			fprintf(stderr, "\\%03o", c);
		else
			fprintf(stderr, "%c", c);
}

xsdotc()
{
	register FILE *strf = fopen(strings, "r");
	register FILE *xdotcf;

	if (strf == NULL)
		perror(strings), exit(5);
	xdotcf = fopen("xs.c", "w");
	if (xdotcf == NULL)
		perror("xs.c"), exit(6);
	fprintf(xdotcf, "char\txstr[] = {\n");
	for (;;) {
		register int i, c;

		for (i = 0; i < 8; i++) {
			c = getc(strf);
			if (ferror(strf)) {
				perror(strings);
				onintr();
			}
			if (feof(strf)) {
				fprintf(xdotcf, "\n");
				goto out;
			}
			fprintf(xdotcf, "0x%02x,", c);
		}
		fprintf(xdotcf, "\n");
	}
out:
	fprintf(xdotcf, "};\n");
	ignore(fclose(xdotcf));
	ignore(fclose(strf));
}

lastchr(cp)
	register char *cp;
{

	while (cp[0] && cp[1])
		cp++;
	return (*cp);
}

istail(str, of)
	register char *str, *of;
{
	register int d = strlen(of) - strlen(str);

	if (d < 0 || strcmp(&of[d], str) != 0)
		return (-1);
	return (d);
}

void
onintr()
{

	ignore(signal(SIGINT, SIG_IGN));
	if (strings[0] == '/')
		ignore(unlink(strings));
	ignore(unlink("x.c"));
	ignore(unlink("xs.c"));
	exit(7);
}
Commit	Line	Data
f42904bc	1	/*
fa6237e9 KB	2	* Copyright (c) 1980 The Regents of the University of California.
	3	* All rights reserved.
	4	*
6ecf3d85	5	* %sccs.include.redist.c%
f42904bc DF	6	*/
	7
	8	#ifndef lint
	9	char copyright[] =
fa6237e9	10	"@(#) Copyright (c) 1980 The Regents of the University of California.\n\
f42904bc	11	All rights reserved.\n";
fa6237e9	12	#endif /* not lint */
f42904bc DF	13
f42904bc DF	14	#ifndef lint
976a4efb	15	static char sccsid[] = "@(#)xstr.c 5.7 (Berkeley) %G%";
fa6237e9	16	#endif /* not lint */
f42904bc	17
4db42f7f BJ	18	#include <sys/types.h>
4db42f7f BJ	19	#include <signal.h>
976a4efb KB	20	#include <errno.h>
976a4efb KB	21	#include <unistd.h>
dc0e9d50 KB	22	#include <stdio.h>
dc0e9d50 KB	23	#include <ctype.h>
976a4efb	24	#include <string.h>
dc0e9d50	25	#include "pathnames.h"
4db42f7f BJ	26
	27	/*
	28	* xstr - extract and hash strings in a C program
	29	*
	30	* Bill Joy UCB
	31	* November, 1978
	32	*/
	33
654d8cfc	34	#define ignore(a) ((void) a)
4db42f7f	35
4db42f7f BJ	36	off_t tellpt;
4db42f7f BJ	37	off_t hashit();
976a4efb	38	void onintr();
4db42f7f	39	char *savestr();
4db42f7f BJ	40	off_t yankstr();
	41
	42	off_t mesgpt;
	43	char *strings = "strings";
	44
	45	int cflg;
	46	int vflg;
	47	int readstd;
	48
	49	main(argc, argv)
	50	int argc;
	51	char *argv[];
	52	{
	53
	54	argc--, argv++;
	55	while (argc > 0 && argv[0][0] == '-') {
	56	register char cp = &(argv++)[1];
	57
	58	argc--;
	59	if (*cp == 0) {
	60	readstd++;
	61	continue;
	62	}
	63	do switch (*cp++) {
	64
	65	case 'c':
	66	cflg++;
	67	continue;
	68
	69	case 'v':
	70	vflg++;
	71	continue;
	72
	73	default:
	74	fprintf(stderr, "usage: xstr [ -v ] [ -c ] [ - ] [ name ... ]\n");
	75	} while (*cp);
	76	}
	77	if (signal(SIGINT, SIG_IGN) == SIG_DFL)
	78	signal(SIGINT, onintr);
	79	if (cflg \|\| argc == 0 && !readstd)
	80	inithash();
	81	else
976a4efb	82	strings = mktemp(strdup(_PATH_TMP));
4db42f7f BJ	83	while (readstd \|\| argc > 0) {
	84	if (freopen("x.c", "w", stdout) == NULL)
	85	perror("x.c"), exit(1);
	86	if (!readstd && freopen(argv[0], "r", stdin) == NULL)
	87	perror(argv[0]), exit(2);
	88	process("x.c");
	89	if (readstd == 0)
	90	argc--, argv++;
	91	else
	92	readstd = 0;
	93	};
	94	flushsh();
	95	if (cflg == 0)
	96	xsdotc();
	97	if (strings[0] == '/')
	98	ignore(unlink(strings));
	99	exit(0);
	100	}
	101
48b802d7 JB	102	char linebuf[BUFSIZ];
48b802d7 JB	103
4db42f7f BJ	104	process(name)
	105	char *name;
	106	{
	107	char *cp;
4db42f7f BJ	108	register int c;
4db42f7f BJ	109	register int incomm = 0;
48b802d7	110	int ret;
4db42f7f	111
48b802d7	112	printf("extern char\txstr[];\n");
4db42f7f BJ	113	for (;;) {
	114	if (fgets(linebuf, sizeof linebuf, stdin) == NULL) {
	115	if (ferror(stdin)) {
	116	perror(name);
	117	exit(3);
	118	}
	119	break;
	120	}
	121	if (linebuf[0] == '#') {
	122	if (linebuf[1] == ' ' && isdigit(linebuf[2]))
	123	printf("#line%s", &linebuf[1]);
	124	else
	125	printf("%s", linebuf);
	126	continue;
	127	}
	128	for (cp = linebuf; c = *cp++;) switch (c) {
	129
	130	case '"':
	131	if (incomm)
	132	goto def;
48b802d7 JB	133	if ((ret = (int) yankstr(&cp)) == -1)
	134	goto out;
	135	printf("(&xstr[%d])", ret);
4db42f7f BJ	136	break;
	137
	138	case '\'':
	139	if (incomm)
	140	goto def;
	141	putchar(c);
	142	if (*cp)
	143	putchar(*cp++);
	144	break;
	145
	146	case '/':
	147	if (incomm \|\| cp != '')
	148	goto def;
	149	incomm = 1;
	150	cp++;
	151	printf("/*");
	152	continue;
	153
	154	case '*':
	155	if (incomm && *cp == '/') {
	156	incomm = 0;
	157	cp++;
	158	printf("*/");
	159	continue;
	160	}
	161	goto def;
	162
	163	def:
	164	default:
	165	putchar(c);
	166	break;
	167	}
	168	}
48b802d7	169	out:
4db42f7f BJ	170	if (ferror(stdout))
	171	perror("x.c"), onintr();
	172	}
	173
	174	off_t
	175	yankstr(cpp)
	176	register char **cpp;
	177	{
	178	register char cp = cpp;
	179	register int c, ch;
	180	char dbuf[BUFSIZ];
	181	register char *dp = dbuf;
	182	register char *tp;
	183
	184	while (c = *cp++) {
	185	switch (c) {
	186
	187	case '"':
	188	cp++;
	189	goto out;
	190
	191	case '\\':
	192	c = *cp++;
	193	if (c == 0)
	194	break;
48b802d7 JB	195	if (c == '\n') {
	196	if (fgets(linebuf, sizeof linebuf, stdin)
	197	== NULL) {
	198	if (ferror(stdin)) {
	199	perror("x.c");
	200	exit(3);
	201	}
	202	return(-1);
	203	}
	204	cp = linebuf;
4db42f7f	205	continue;
48b802d7	206	}
4db42f7f BJ	207	for (tp = "b\bt\tr\rn\nf\f\\\\\"\""; ch = *tp++; tp++)
	208	if (c == ch) {
	209	c = *tp;
	210	goto gotc;
	211	}
	212	if (!octdigit(c)) {
	213	*dp++ = '\\';
	214	break;
	215	}
	216	c -= '0';
	217	if (!octdigit(*cp))
	218	break;
	219	c <<= 3, c += *cp++ - '0';
	220	if (!octdigit(*cp))
	221	break;
	222	c <<= 3, c += *cp++ - '0';
	223	break;
	224	}
	225	gotc:
	226	*dp++ = c;
	227	}
	228	out:
	229	*cpp = --cp;
	230	*dp = 0;
	231	return (hashit(dbuf, 1));
	232	}
	233
	234	octdigit(c)
	235	char c;
	236	{
	237
	238	return (isdigit(c) && c != '8' && c != '9');
	239	}
	240
	241	inithash()
	242	{
	243	char buf[BUFSIZ];
	244	register FILE *mesgread = fopen(strings, "r");
	245
	246	if (mesgread == NULL)
	247	return;
	248	for (;;) {
	249	mesgpt = tellpt;
	250	if (fgetNUL(buf, sizeof buf, mesgread) == NULL)
	251	break;
	252	ignore(hashit(buf, 0));
	253	}
	254	ignore(fclose(mesgread));
	255	}
	256
	257	fgetNUL(obuf, rmdr, file)
	258	char *obuf;
	259	register int rmdr;
	260	FILE *file;
	261	{
	262	register c;
	263	register char *buf = obuf;
	264
	265	while (--rmdr > 0 && (c = xgetc(file)) != 0 && c != EOF)
	266	*buf++ = c;
	267	*buf++ = 0;
	268	return ((feof(file) \|\| ferror(file)) ? NULL : 1);
	269	}
	270
271	xgetc(file)
272	FILE *file;
273	{
274
275	tellpt++;
276	return (getc(file));
277	}
278
279	#define BUCKETS 128
280
281	struct hash {
282	off_t hpt;
283	char *hstr;
284	struct hash *hnext;
285	short hnew;
286	} bucket[BUCKETS];
287
288	off_t
289	hashit(str, new)
290	char *str;
291	int new;
292	{
293	int i;
294	register struct hash hp, hp0;
295
296	hp = hp0 = &bucket[lastchr(str) & 0177];
297	while (hp->hnext) {
298	hp = hp->hnext;
299	i = istail(str, hp->hstr);
300	if (i >= 0)
301	return (hp->hpt + i);
302	}
654d8cfc KM	303	if ((hp = (struct hash ) calloc(1, sizeof (hp))) == NULL) {
	304	perror("xstr");
	305	exit(8);
	306	}
4db42f7f	307	hp->hpt = mesgpt;
976a4efb KB	308	if (!(hp->hstr = strdup(str))) {
	309	(void)fprintf(stderr, "xstr: %s\n", strerror(errno));
	310	exit(1);
	311	}
4db42f7f BJ	312	mesgpt += strlen(hp->hstr) + 1;
	313	hp->hnext = hp0->hnext;
	314	hp->hnew = new;
	315	hp0->hnext = hp;
	316	return (hp->hpt);
	317	}
	318
	319	flushsh()
	320	{
	321	register int i;
	322	register struct hash *hp;
	323	register FILE *mesgwrit;
	324	register int old = 0, new = 0;
	325
	326	for (i = 0; i < BUCKETS; i++)
	327	for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext)
	328	if (hp->hnew)
	329	new++;
	330	else
	331	old++;
	332	if (new == 0 && old != 0)
	333	return;
dd558f1f	334	mesgwrit = fopen(strings, old ? "r+" : "w");
f5affc92 BJ	335	if (mesgwrit == NULL)
f5affc92 BJ	336	perror(strings), exit(4);
4db42f7f BJ	337	for (i = 0; i < BUCKETS; i++)
	338	for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext) {
	339	found(hp->hnew, hp->hpt, hp->hstr);
	340	if (hp->hnew) {
	341	fseek(mesgwrit, hp->hpt, 0);
	342	ignore(fwrite(hp->hstr, strlen(hp->hstr) + 1, 1, mesgwrit));
	343	if (ferror(mesgwrit))
	344	perror(strings), exit(4);
	345	}
	346	}
654d8cfc KM	347	if (fclose(mesgwrit) == EOF)
654d8cfc KM	348	perror(strings), exit(4);
4db42f7f BJ	349	}
	350
	351	found(new, off, str)
	352	int new;
	353	off_t off;
	354	char *str;
	355	{
4db42f7f BJ	356	if (vflg == 0)
	357	return;
	358	if (!new)
	359	fprintf(stderr, "found at %d:", (int) off);
	360	else
	361	fprintf(stderr, "new at %d:", (int) off);
	362	prstr(str);
	363	fprintf(stderr, "\n");
	364	}
	365
	366	prstr(cp)
	367	register char *cp;
	368	{
	369	register int c;
	370
	371	while (c = (*cp++ & 0377))
	372	if (c < ' ')
	373	fprintf(stderr, "^%c", c + '`');
	374	else if (c == 0177)
	375	fprintf(stderr, "^?");
	376	else if (c > 0200)
	377	fprintf(stderr, "\\%03o", c);
	378	else
	379	fprintf(stderr, "%c", c);
	380	}
	381
	382	xsdotc()
	383	{
	384	register FILE *strf = fopen(strings, "r");
	385	register FILE *xdotcf;
	386
	387	if (strf == NULL)
	388	perror(strings), exit(5);
	389	xdotcf = fopen("xs.c", "w");
	390	if (xdotcf == NULL)
	391	perror("xs.c"), exit(6);
	392	fprintf(xdotcf, "char\txstr[] = {\n");
	393	for (;;) {
	394	register int i, c;
	395
	396	for (i = 0; i < 8; i++) {
	397	c = getc(strf);
	398	if (ferror(strf)) {
	399	perror(strings);
	400	onintr();
	401	}
	402	if (feof(strf)) {
	403	fprintf(xdotcf, "\n");
	404	goto out;
	405	}
	406	fprintf(xdotcf, "0x%02x,", c);
	407	}
	408	fprintf(xdotcf, "\n");
	409	}
	410	out:
	411	fprintf(xdotcf, "};\n");
	412	ignore(fclose(xdotcf));
	413	ignore(fclose(strf));
	414	}
	415
4db42f7f BJ	416	lastchr(cp)
	417	register char *cp;
	418	{
	419
	420	while (cp[0] && cp[1])
	421	cp++;
	422	return (*cp);
	423	}
	424
	425	istail(str, of)
	426	register char str, of;
	427	{
	428	register int d = strlen(of) - strlen(str);
	429
	430	if (d < 0 \|\| strcmp(&of[d], str) != 0)
	431	return (-1);
	432	return (d);
	433	}
	434
976a4efb	435	void
4db42f7f BJ	436	onintr()
	437	{
	438
654d8cfc	439	ignore(signal(SIGINT, SIG_IGN));
4db42f7f BJ	440	if (strings[0] == '/')
	441	ignore(unlink(strings));
	442	ignore(unlink("x.c"));
	443	ignore(unlink("xs.c"));
	444	exit(7);
	445	}