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

#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)	Ignore((char *) 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);
}

process(name)
	char *name;
{
	char *cp;
	char linebuf[BUFSIZ];
	register int c;
	register int incomm = 0;

	printf("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;
			printf("(&xstr[%d])", (int) yankstr(&cp));
			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;
		}
	}
	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')
				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);
	}
	hp = (struct hash *) calloc(1, sizeof (*hp));
	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 ? "a" : "w");
	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);
			}
		}
	ignore(fclose(mesgwrit));
}

found(new, off, str)
	int new;
	off_t off;
	char *str;
{
	register char *cp;

	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 = (char *) calloc(1, strlen(cp) + 1);

	return (strcpy(dp, cp));
}

Ignore(a)
	char *a;
{

	a = a;
}

ignorf(a)
	int (*a)();
{

	a = a;
}

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()
{

	ignorf(signal(SIGINT, SIG_IGN));
	if (strings[0] == '/')
		ignore(unlink(strings));
	ignore(unlink("x.c"));
	ignore(unlink("xs.c"));
	exit(7);
}
Commit	Line	Data
892a661d BJ	1	#include <stdio.h>
	2	#include <ctype.h>
	3	#include <sys/types.h>
	4	#include <signal.h>
	5
	6	/*
	7	* xstr - extract and hash strings in a C program
	8	*
	9	* Bill Joy UCB
	10	* November, 1978
	11	*/
	12
	13	#define ignore(a) Ignore((char *) a)
	14
	15	char *calloc();
	16	off_t tellpt;
	17	off_t hashit();
	18	char *mktemp();
	19	int onintr();
	20	char *savestr();
	21	char *strcat();
	22	char *strcpy();
	23	off_t yankstr();
	24
	25	off_t mesgpt;
	26	char *strings = "strings";
	27
	28	int cflg;
	29	int vflg;
	30	int readstd;
	31
	32	main(argc, argv)
	33	int argc;
	34	char *argv[];
	35	{
	36
	37	argc--, argv++;
	38	while (argc > 0 && argv[0][0] == '-') {
	39	register char cp = &(argv++)[1];
	40
	41	argc--;
	42	if (*cp == 0) {
	43	readstd++;
	44	continue;
	45	}
	46	do switch (*cp++) {
	47
	48	case 'c':
	49	cflg++;
	50	continue;
	51
	52	case 'v':
	53	vflg++;
	54	continue;
	55
	56	default:
	57	fprintf(stderr, "usage: xstr [ -v ] [ -c ] [ - ] [ name ... ]\n");
	58	} while (*cp);
	59	}
	60	if (signal(SIGINT, SIG_IGN) == SIG_DFL)
	61	signal(SIGINT, onintr);
	62	if (cflg \|\| argc == 0 && !readstd)
	63	inithash();
	64	else
65	strings = mktemp(savestr("/tmp/xstrXXXXXX"));
66	while (readstd \|\| argc > 0) {
67	if (freopen("x.c", "w", stdout) == NULL)
68	perror("x.c"), exit(1);
69	if (!readstd && freopen(argv[0], "r", stdin) == NULL)
70	perror(argv[0]), exit(2);
71	process("x.c");
72	if (readstd == 0)
73	argc--, argv++;
74	else
75	readstd = 0;
76	};
77	flushsh();
78	if (cflg == 0)
79	xsdotc();
80	if (strings[0] == '/')
81	ignore(unlink(strings));
82	exit(0);
83	}
84
85	process(name)
86	char *name;
87	{
88	char *cp;
89	char linebuf[BUFSIZ];
90	register int c;
91	register int incomm = 0;
92
93	printf("char\txstr[];\n");
94	for (;;) {
95	if (fgets(linebuf, sizeof linebuf, stdin) == NULL) {
96	if (ferror(stdin)) {
97	perror(name);
98	exit(3);
99	}
100	break;
101	}
102	if (linebuf[0] == '#') {
103	if (linebuf[1] == ' ' && isdigit(linebuf[2]))
104	printf("#line%s", &linebuf[1]);
105	else
106	printf("%s", linebuf);
107	continue;
108	}
109	for (cp = linebuf; c = *cp++;) switch (c) {
110
111	case '"':
112	if (incomm)
113	goto def;
114	printf("(&xstr[%d])", (int) yankstr(&cp));
115	break;
116
117	case '\'':
118	if (incomm)
119	goto def;
120	putchar(c);
121	if (*cp)
122	putchar(*cp++);
123	break;
124
125	case '/':
126	if (incomm \|\| cp != '')
127	goto def;
128	incomm = 1;
129	cp++;
130	printf("/*");
131	continue;
132
133	case '*':
134	if (incomm && *cp == '/') {
135	incomm = 0;
136	cp++;
137	printf("*/");
138	continue;
139	}
140	goto def;
141
142	def:
143	default:
144	putchar(c);
145	break;
146	}
147	}
148	if (ferror(stdout))
149	perror("x.c"), onintr();
150	}
151
152	off_t
153	yankstr(cpp)
154	register char **cpp;
155	{
156	register char cp = cpp;
157	register int c, ch;
158	char dbuf[BUFSIZ];
159	register char *dp = dbuf;
160	register char *tp;
161
162	while (c = *cp++) {
163	switch (c) {
164
165	case '"':
166	cp++;
167	goto out;
168
169	case '\\':
170	c = *cp++;
171	if (c == 0)
172	break;
173	if (c == '\n')
174	continue;
175	for (tp = "b\bt\tr\rn\nf\f\\\\\"\""; ch = *tp++; tp++)
176	if (c == ch) {
177	c = *tp;
178	goto gotc;
179	}
180	if (!octdigit(c)) {
181	*dp++ = '\\';
182	break;
183	}
184	c -= '0';
185	if (!octdigit(*cp))
186	break;
187	c <<= 3, c += *cp++ - '0';
188	if (!octdigit(*cp))
189	break;
190	c <<= 3, c += *cp++ - '0';
191	break;
192	}
193	gotc:
194	*dp++ = c;
195	}
196	out:
197	*cpp = --cp;
198	*dp = 0;
199	return (hashit(dbuf, 1));
200	}
201
202	octdigit(c)
203	char c;
204	{
205
206	return (isdigit(c) && c != '8' && c != '9');
207	}
208
209	inithash()
210	{
211	char buf[BUFSIZ];
212	register FILE *mesgread = fopen(strings, "r");
213
214	if (mesgread == NULL)
215	return;
216	for (;;) {
217	mesgpt = tellpt;
218	if (fgetNUL(buf, sizeof buf, mesgread) == NULL)
219	break;
220	ignore(hashit(buf, 0));
221	}
222	ignore(fclose(mesgread));
223	}
224
225	fgetNUL(obuf, rmdr, file)
226	char *obuf;
227	register int rmdr;
228	FILE *file;
229	{
230	register c;
231	register char *buf = obuf;
232
233	while (--rmdr > 0 && (c = xgetc(file)) != 0 && c != EOF)
234	*buf++ = c;
235	*buf++ = 0;
236	return ((feof(file) \|\| ferror(file)) ? NULL : 1);
237	}
238
239	xgetc(file)
240	FILE *file;
241	{
242
243	tellpt++;
244	return (getc(file));
245	}
246
247	#define BUCKETS 128
248
249	struct hash {
250	off_t hpt;
251	char *hstr;
252	struct hash *hnext;
253	short hnew;
254	} bucket[BUCKETS];
255
256	off_t
257	hashit(str, new)
258	char *str;
259	int new;
260	{
261	int i;
262	register struct hash hp, hp0;
263
264	hp = hp0 = &bucket[lastchr(str) & 0177];
265	while (hp->hnext) {
266	hp = hp->hnext;
267	i = istail(str, hp->hstr);
268	if (i >= 0)
269	return (hp->hpt + i);
270	}
271	hp = (struct hash ) calloc(1, sizeof (hp));
272	hp->hpt = mesgpt;
273	hp->hstr = savestr(str);
274	mesgpt += strlen(hp->hstr) + 1;
275	hp->hnext = hp0->hnext;
276	hp->hnew = new;
277	hp0->hnext = hp;
278	return (hp->hpt);
279	}
280
281	flushsh()
282	{
283	register int i;
284	register struct hash *hp;
285	register FILE *mesgwrit;
286	register int old = 0, new = 0;
287
288	for (i = 0; i < BUCKETS; i++)
289	for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext)
290	if (hp->hnew)
291	new++;
292	else
293	old++;
294	if (new == 0 && old != 0)
295	return;
296	mesgwrit = fopen(strings, old ? "a" : "w");
297	for (i = 0; i < BUCKETS; i++)
298	for (hp = bucket[i].hnext; hp != NULL; hp = hp->hnext) {
299	found(hp->hnew, hp->hpt, hp->hstr);
300	if (hp->hnew) {
301	fseek(mesgwrit, hp->hpt, 0);
302	ignore(fwrite(hp->hstr, strlen(hp->hstr) + 1, 1, mesgwrit));
303	if (ferror(mesgwrit))
304	perror(strings), exit(4);
305	}
306	}
307	ignore(fclose(mesgwrit));
308	}
309
310	found(new, off, str)
311	int new;
312	off_t off;
313	char *str;
314	{
315	register char *cp;
316
317	if (vflg == 0)
318	return;
319	if (!new)
320	fprintf(stderr, "found at %d:", (int) off);
321	else
322	fprintf(stderr, "new at %d:", (int) off);
323	prstr(str);
324	fprintf(stderr, "\n");
325	}
326
327	prstr(cp)
328	register char *cp;
329	{
330	register int c;
331
332	while (c = (*cp++ & 0377))
333	if (c < ' ')
334	fprintf(stderr, "^%c", c + '`');
335	else if (c == 0177)
336	fprintf(stderr, "^?");
337	else if (c > 0200)
338	fprintf(stderr, "\\%03o", c);
339	else
340	fprintf(stderr, "%c", c);
341	}
342
343	xsdotc()
344	{
345	register FILE *strf = fopen(strings, "r");
346	register FILE *xdotcf;
347
348	if (strf == NULL)
349	perror(strings), exit(5);
350	xdotcf = fopen("xs.c", "w");
351	if (xdotcf == NULL)
352	perror("xs.c"), exit(6);
353	fprintf(xdotcf, "char\txstr[] = {\n");
354	for (;;) {
355	register int i, c;
356
357	for (i = 0; i < 8; i++) {
358	c = getc(strf);
359	if (ferror(strf)) {
360	perror(strings);
361	onintr();
362	}
363	if (feof(strf)) {
364	fprintf(xdotcf, "\n");
365	goto out;
366	}
367	fprintf(xdotcf, "0x%02x,", c);
368	}
369	fprintf(xdotcf, "\n");
370	}
371	out:
372	fprintf(xdotcf, "};\n");
373	ignore(fclose(xdotcf));
374	ignore(fclose(strf));
375	}
376
377	char *
378	savestr(cp)
379	register char *cp;
380	{
381	register char dp = (char ) calloc(1, strlen(cp) + 1);
382
383	return (strcpy(dp, cp));
384	}
385
386	Ignore(a)
387	char *a;
388	{
389
390	a = a;
391	}
392
393	ignorf(a)
394	int (*a)();
395	{
396
397	a = a;
398	}
399
400	lastchr(cp)
401	register char *cp;
402	{
403
404	while (cp[0] && cp[1])
405	cp++;
406	return (*cp);
407	}
408
409	istail(str, of)
410	register char str, of;
411	{
412	register int d = strlen(of) - strlen(str);
413
414	if (d < 0 \|\| strcmp(&of[d], str) != 0)
415	return (-1);
416	return (d);
417	}
418
419	onintr()
420	{
421
422	ignorf(signal(SIGINT, SIG_IGN));
423	if (strings[0] == '/')
424	ignore(unlink(strings));
425	ignore(unlink("x.c"));
426	ignore(unlink("xs.c"));
427	exit(7);
428	}