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

/*
 * Copyright (c) 1980 Regents of the University of California.
 * All rights reserved.  The Berkeley software License Agreement
 * specifies the terms and conditions for redistribution.
 */

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

#ifndef lint
static char sccsid[] = "@(#)xstr.c	5.4 (Berkeley) %G%";
#endif not lint

#include <stdio.h>
#include <ctype.h>
#include <sys/types.h>
#include <signal.h>

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

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

char	*calloc();
off_t	tellpt;
off_t	hashit();
char	*mktemp();
int	onintr();
char	*savestr();
char	*strcat();
char	*strcpy();
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(savestr("/tmp/xstrXXXXXX"));
	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;
	hp->hstr = savestr(str);
	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));
}

char *
savestr(cp)
	register char *cp;
{
	register char *dp;

	if ((dp = (char *) calloc(1, strlen(cp) + 1)) == NULL) {
		perror("xstr");
		exit(8);
	}
	return (strcpy(dp, cp));
}

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);
}

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 DF	1	/*
	2	* Copyright (c) 1980 Regents of the University of California.
	3	* All rights reserved. The Berkeley software License Agreement
	4	* specifies the terms and conditions for redistribution.
	5	*/
	6
	7	#ifndef lint
	8	char copyright[] =
	9	"@(#) Copyright (c) 1980 Regents of the University of California.\n\
	10	All rights reserved.\n";
	11	#endif not lint
	12
	13	#ifndef lint
40b83fad	14	static char sccsid[] = "@(#)xstr.c 5.4 (Berkeley) %G%";
f42904bc DF	15	#endif not lint
f42904bc DF	16
4db42f7f BJ	17	#include <stdio.h>
	18	#include <ctype.h>
	19	#include <sys/types.h>
	20	#include <signal.h>
	21
	22	/*
	23	* xstr - extract and hash strings in a C program
	24	*
	25	* Bill Joy UCB
	26	* November, 1978
	27	*/
	28
654d8cfc	29	#define ignore(a) ((void) a)
4db42f7f BJ	30
	31	char *calloc();
	32	off_t tellpt;
	33	off_t hashit();
	34	char *mktemp();
	35	int onintr();
	36	char *savestr();
	37	char *strcat();
	38	char *strcpy();
	39	off_t yankstr();
	40
	41	off_t mesgpt;
	42	char *strings = "strings";
	43
	44	int cflg;
	45	int vflg;
	46	int readstd;
	47
	48	main(argc, argv)
	49	int argc;
	50	char *argv[];
	51	{
	52
	53	argc--, argv++;
	54	while (argc > 0 && argv[0][0] == '-') {
	55	register char cp = &(argv++)[1];
	56
	57	argc--;
	58	if (*cp == 0) {
	59	readstd++;
	60	continue;
	61	}
	62	do switch (*cp++) {
	63
	64	case 'c':
	65	cflg++;
	66	continue;
	67
	68	case 'v':
	69	vflg++;
	70	continue;
	71
	72	default:
	73	fprintf(stderr, "usage: xstr [ -v ] [ -c ] [ - ] [ name ... ]\n");
	74	} while (*cp);
	75	}
	76	if (signal(SIGINT, SIG_IGN) == SIG_DFL)
	77	signal(SIGINT, onintr);
	78	if (cflg \|\| argc == 0 && !readstd)
	79	inithash();
	80	else
	81	strings = mktemp(savestr("/tmp/xstrXXXXXX"));
	82	while (readstd \|\| argc > 0) {
	83	if (freopen("x.c", "w", stdout) == NULL)
	84	perror("x.c"), exit(1);
	85	if (!readstd && freopen(argv[0], "r", stdin) == NULL)
	86	perror(argv[0]), exit(2);
	87	process("x.c");
	88	if (readstd == 0)
	89	argc--, argv++;
	90	else
	91	readstd = 0;
	92	};
	93	flushsh();
94	if (cflg == 0)
95	xsdotc();
96	if (strings[0] == '/')
97	ignore(unlink(strings));
98	exit(0);
99	}
100
48b802d7 JB	101	char linebuf[BUFSIZ];
48b802d7 JB	102
4db42f7f BJ	103	process(name)
	104	char *name;
	105	{
	106	char *cp;
4db42f7f BJ	107	register int c;
4db42f7f BJ	108	register int incomm = 0;
48b802d7	109	int ret;
4db42f7f	110
48b802d7	111	printf("extern char\txstr[];\n");
4db42f7f BJ	112	for (;;) {
	113	if (fgets(linebuf, sizeof linebuf, stdin) == NULL) {
	114	if (ferror(stdin)) {
	115	perror(name);
	116	exit(3);
	117	}
	118	break;
	119	}
	120	if (linebuf[0] == '#') {
	121	if (linebuf[1] == ' ' && isdigit(linebuf[2]))
	122	printf("#line%s", &linebuf[1]);
	123	else
	124	printf("%s", linebuf);
	125	continue;
	126	}
	127	for (cp = linebuf; c = *cp++;) switch (c) {
	128
	129	case '"':
	130	if (incomm)
	131	goto def;
48b802d7 JB	132	if ((ret = (int) yankstr(&cp)) == -1)
	133	goto out;
	134	printf("(&xstr[%d])", ret);
4db42f7f BJ	135	break;
	136
	137	case '\'':
	138	if (incomm)
	139	goto def;
	140	putchar(c);
	141	if (*cp)
	142	putchar(*cp++);
	143	break;
	144
	145	case '/':
	146	if (incomm \|\| cp != '')
	147	goto def;
	148	incomm = 1;
	149	cp++;
	150	printf("/*");
	151	continue;
	152
	153	case '*':
	154	if (incomm && *cp == '/') {
	155	incomm = 0;
	156	cp++;
	157	printf("*/");
	158	continue;
	159	}
	160	goto def;
	161
	162	def:
	163	default:
	164	putchar(c);
	165	break;
	166	}
	167	}
48b802d7	168	out:
4db42f7f BJ	169	if (ferror(stdout))
	170	perror("x.c"), onintr();
	171	}
	172
	173	off_t
	174	yankstr(cpp)
	175	register char **cpp;
	176	{
	177	register char cp = cpp;
	178	register int c, ch;
	179	char dbuf[BUFSIZ];
	180	register char *dp = dbuf;
	181	register char *tp;
	182
	183	while (c = *cp++) {
	184	switch (c) {
	185
	186	case '"':
	187	cp++;
	188	goto out;
	189
	190	case '\\':
	191	c = *cp++;
	192	if (c == 0)
	193	break;
48b802d7 JB	194	if (c == '\n') {
	195	if (fgets(linebuf, sizeof linebuf, stdin)
	196	== NULL) {
	197	if (ferror(stdin)) {
	198	perror("x.c");
	199	exit(3);
	200	}
	201	return(-1);
	202	}
	203	cp = linebuf;
4db42f7f	204	continue;
48b802d7	205	}
4db42f7f BJ	206	for (tp = "b\bt\tr\rn\nf\f\\\\\"\""; ch = *tp++; tp++)
	207	if (c == ch) {
	208	c = *tp;
	209	goto gotc;
	210	}
	211	if (!octdigit(c)) {
	212	*dp++ = '\\';
	213	break;
	214	}
	215	c -= '0';
	216	if (!octdigit(*cp))
	217	break;
	218	c <<= 3, c += *cp++ - '0';
	219	if (!octdigit(*cp))
	220	break;
	221	c <<= 3, c += *cp++ - '0';
	222	break;
	223	}
	224	gotc:
	225	*dp++ = c;
	226	}
	227	out:
	228	*cpp = --cp;
	229	*dp = 0;
	230	return (hashit(dbuf, 1));
	231	}
	232
	233	octdigit(c)
	234	char c;
	235	{
	236
	237	return (isdigit(c) && c != '8' && c != '9');
	238	}
	239
	240	inithash()
	241	{
	242	char buf[BUFSIZ];
	243	register FILE *mesgread = fopen(strings, "r");
	244
	245	if (mesgread == NULL)
	246	return;
	247	for (;;) {
	248	mesgpt = tellpt;
	249	if (fgetNUL(buf, sizeof buf, mesgread) == NULL)
	250	break;
	251	ignore(hashit(buf, 0));
	252	}
	253	ignore(fclose(mesgread));
	254	}
	255
	256	fgetNUL(obuf, rmdr, file)
	257	char *obuf;
	258	register int rmdr;
	259	FILE *file;
	260	{
	261	register c;
	262	register char *buf = obuf;
	263
	264	while (--rmdr > 0 && (c = xgetc(file)) != 0 && c != EOF)
	265	*buf++ = c;
	266	*buf++ = 0;
	267	return ((feof(file) \|\| ferror(file)) ? NULL : 1);
	268	}
	269
270	xgetc(file)
271	FILE *file;
272	{
273
274	tellpt++;
275	return (getc(file));
276	}
277
278	#define BUCKETS 128
279
280	struct hash {
281	off_t hpt;
282	char *hstr;
283	struct hash *hnext;
284	short hnew;
285	} bucket[BUCKETS];
286
287	off_t
288	hashit(str, new)
289	char *str;
290	int new;
291	{
292	int i;
293	register struct hash hp, hp0;
294
295	hp = hp0 = &bucket[lastchr(str) & 0177];
296	while (hp->hnext) {
297	hp = hp->hnext;
298	i = istail(str, hp->hstr);
299	if (i >= 0)
300	return (hp->hpt + i);
301	}
654d8cfc KM	302	if ((hp = (struct hash ) calloc(1, sizeof (hp))) == NULL) {
	303	perror("xstr");
	304	exit(8);
	305	}
4db42f7f BJ	306	hp->hpt = mesgpt;
	307	hp->hstr = savestr(str);
	308	mesgpt += strlen(hp->hstr) + 1;
	309	hp->hnext = hp0->hnext;
	310	hp->hnew = new;
	311	hp0->hnext = hp;
	312	return (hp->hpt);
	313	}
	314
	315	flushsh()
	316	{
	317	register int i;
	318	register struct hash *hp;
	319	register FILE *mesgwrit;
	320	register int old = 0, new = 0;
	321
	322	for (i = 0; i < BUCKETS; i++)
	323	for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext)
	324	if (hp->hnew)
	325	new++;
	326	else
	327	old++;
	328	if (new == 0 && old != 0)
	329	return;
dd558f1f	330	mesgwrit = fopen(strings, old ? "r+" : "w");
f5affc92 BJ	331	if (mesgwrit == NULL)
f5affc92 BJ	332	perror(strings), exit(4);
4db42f7f BJ	333	for (i = 0; i < BUCKETS; i++)
	334	for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext) {
	335	found(hp->hnew, hp->hpt, hp->hstr);
	336	if (hp->hnew) {
	337	fseek(mesgwrit, hp->hpt, 0);
	338	ignore(fwrite(hp->hstr, strlen(hp->hstr) + 1, 1, mesgwrit));
	339	if (ferror(mesgwrit))
	340	perror(strings), exit(4);
	341	}
	342	}
654d8cfc KM	343	if (fclose(mesgwrit) == EOF)
654d8cfc KM	344	perror(strings), exit(4);
4db42f7f BJ	345	}
	346
	347	found(new, off, str)
	348	int new;
	349	off_t off;
	350	char *str;
	351	{
4db42f7f BJ	352	if (vflg == 0)
	353	return;
	354	if (!new)
	355	fprintf(stderr, "found at %d:", (int) off);
	356	else
	357	fprintf(stderr, "new at %d:", (int) off);
	358	prstr(str);
	359	fprintf(stderr, "\n");
	360	}
	361
	362	prstr(cp)
	363	register char *cp;
	364	{
	365	register int c;
	366
	367	while (c = (*cp++ & 0377))
	368	if (c < ' ')
	369	fprintf(stderr, "^%c", c + '`');
	370	else if (c == 0177)
	371	fprintf(stderr, "^?");
	372	else if (c > 0200)
	373	fprintf(stderr, "\\%03o", c);
	374	else
	375	fprintf(stderr, "%c", c);
	376	}
	377
	378	xsdotc()
	379	{
	380	register FILE *strf = fopen(strings, "r");
	381	register FILE *xdotcf;
	382
	383	if (strf == NULL)
	384	perror(strings), exit(5);
	385	xdotcf = fopen("xs.c", "w");
	386	if (xdotcf == NULL)
	387	perror("xs.c"), exit(6);
	388	fprintf(xdotcf, "char\txstr[] = {\n");
	389	for (;;) {
	390	register int i, c;
	391
	392	for (i = 0; i < 8; i++) {
	393	c = getc(strf);
	394	if (ferror(strf)) {
	395	perror(strings);
	396	onintr();
	397	}
	398	if (feof(strf)) {
	399	fprintf(xdotcf, "\n");
	400	goto out;
	401	}
	402	fprintf(xdotcf, "0x%02x,", c);
	403	}
	404	fprintf(xdotcf, "\n");
	405	}
	406	out:
	407	fprintf(xdotcf, "};\n");
	408	ignore(fclose(xdotcf));
	409	ignore(fclose(strf));
	410	}
	411
	412	char *
	413	savestr(cp)
	414	register char *cp;
	415	{
654d8cfc	416	register char *dp;
4db42f7f	417
654d8cfc KM	418	if ((dp = (char *) calloc(1, strlen(cp) + 1)) == NULL) {
	419	perror("xstr");
	420	exit(8);
	421	}
4db42f7f BJ	422	return (strcpy(dp, cp));
	423	}
	424
4db42f7f BJ	425	lastchr(cp)
	426	register char *cp;
	427	{
	428
	429	while (cp[0] && cp[1])
	430	cp++;
	431	return (*cp);
	432	}
	433
	434	istail(str, of)
	435	register char str, of;
	436	{
	437	register int d = strlen(of) - strlen(str);
	438
	439	if (d < 0 \|\| strcmp(&of[d], str) != 0)
	440	return (-1);
	441	return (d);
	442	}
	443
	444	onintr()
	445	{
	446
654d8cfc	447	ignore(signal(SIGINT, SIG_IGN));
4db42f7f BJ	448	if (strings[0] == '/')
	449	ignore(unlink(strings));
	450	ignore(unlink("x.c"));
	451	ignore(unlink("xs.c"));
	452	exit(7);
	453	}