xref: /netbsd/usr.bin/m4/gnum4.c (revision 6550d01e) 
1 /* $NetBSD: gnum4.c,v 1.7 2009/10/26 21:16:49 christos Exp $ */
2 /* $OpenBSD: gnum4.c,v 1.39 2008/08/21 21:01:04 espie Exp $ */
3 
4 /*
5  * Copyright (c) 1999 Marc Espie
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  *
16  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
17  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
20  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26  * SUCH DAMAGE.
27  */
28 
29 /*
30  * functions needed to support gnu-m4 extensions, including a fake freezing
31  */
32 #if HAVE_NBTOOL_CONFIG_H
33 #include "nbtool_config.h"
34 #endif
35 #include <sys/cdefs.h>
36 __RCSID("$NetBSD: gnum4.c,v 1.7 2009/10/26 21:16:49 christos Exp $");
37 
38 #include <sys/param.h>
39 #include <sys/types.h>
40 #include <sys/wait.h>
41 #include <ctype.h>
42 #include <err.h>
43 #include <paths.h>
44 #include <regex.h>
45 #include <stddef.h>
46 #include <stdlib.h>
47 #include <stdio.h>
48 #include <string.h>
49 #include <errno.h>
50 #include <unistd.h>
51 #include "mdef.h"
52 #include "stdd.h"
53 #include "extern.h"
54 
55 
56 int mimic_gnu = 0;
57 #ifndef SIZE_T_MAX
58 #define SIZE_T_MAX (size_t)~0ull
59 #endif
60 
61 /*
62  * Support for include path search
63  * First search in the current directory.
64  * If not found, and the path is not absolute, include path kicks in.
65  * First, -I options, in the order found on the command line.
66  * Then M4PATH env variable
67  */
68 
69 struct path_entry {
70 	char *name;
71 	struct path_entry *next;
72 } *first, *last;
73 
74 static struct path_entry *new_path_entry(const char *);
75 static void ensure_m4path(void);
76 static struct input_file *dopath(struct input_file *, const char *);
77 
78 static struct path_entry *
79 new_path_entry(const char *dirname)
80 {
81 	struct path_entry *n;
82 
83 	n = malloc(sizeof(struct path_entry));
84 	if (!n)
85 		errx(1, "out of memory");
86 	n->name = strdup(dirname);
87 	if (!n->name)
88 		errx(1, "out of memory");
89 	n->next = 0;
90 	return n;
91 }
92 
93 void
94 addtoincludepath(const char *dirname)
95 {
96 	struct path_entry *n;
97 
98 	n = new_path_entry(dirname);
99 
100 	if (last) {
101 		last->next = n;
102 		last = n;
103 	}
104 	else
105 		last = first = n;
106 }
107 
108 static void
109 ensure_m4path()
110 {
111 	static int envpathdone = 0;
112 	char *envpath;
113 	char *sweep;
114 	char *path;
115 
116 	if (envpathdone)
117 		return;
118 	envpathdone = TRUE;
119 	envpath = getenv("M4PATH");
120 	if (!envpath)
121 		return;
122 	/* for portability: getenv result is read-only */
123 	envpath = strdup(envpath);
124 	if (!envpath)
125 		errx(1, "out of memory");
126 	for (sweep = envpath;
127 	    (path = strsep(&sweep, ":")) != NULL;)
128 	    addtoincludepath(path);
129 	free(envpath);
130 }
131 
132 static
133 struct input_file *
134 dopath(struct input_file *i, const char *filename)
135 {
136 	char path[MAXPATHLEN];
137 	struct path_entry *pe;
138 	FILE *f;
139 
140 	for (pe = first; pe; pe = pe->next) {
141 		snprintf(path, sizeof(path), "%s/%s", pe->name, filename);
142 		if ((f = fopen(path, "r")) != 0) {
143 			set_input(i, f, path);
144 			return i;
145 		}
146 	}
147 	return NULL;
148 }
149 
150 struct input_file *
151 fopen_trypath(struct input_file *i, const char *filename)
152 {
153 	FILE *f;
154 
155 	f = fopen(filename, "r");
156 	if (f != NULL) {
157 		set_input(i, f, filename);
158 		return i;
159 	}
160 	if (filename[0] == '/')
161 		return NULL;
162 
163 	ensure_m4path();
164 
165 	return dopath(i, filename);
166 }
167 
168 void
169 doindir(const char *argv[], int argc)
170 {
171 	ndptr n;
172 	struct macro_definition *p;
173 
174 	n = lookup(argv[2]);
175 	if (n == NULL || (p = macro_getdef(n)) == NULL)
176 		m4errx(1, "indir: undefined macro %s.", argv[2]);
177 	argv[1] = p->defn;
178 
179 	eval(argv+1, argc-1, p->type, is_traced(n));
180 }
181 
182 void
183 dobuiltin(const char *argv[], int argc)
184 {
185 	ndptr p;
186 
187 	argv[1] = NULL;
188 	p = macro_getbuiltin(argv[2]);
189 	if (p != NULL)
190 		eval(argv+1, argc-1, macro_builtin_type(p), is_traced(p));
191 	else
192 		m4errx(1, "unknown builtin %s.", argv[2]);
193 }
194 
195 
196 /* We need some temporary buffer space, as pb pushes BACK and substitution
197  * proceeds forward... */
198 static char *buffer;
199 static size_t bufsize = 0;
200 static size_t current = 0;
201 
202 static void addchars(const char *, size_t);
203 static void addchar(int);
204 static char *twiddle(const char *);
205 static char *getstring(void);
206 static void exit_regerror(int, regex_t *);
207 static void do_subst(const char *, regex_t *, const char *, regmatch_t *);
208 static void do_regexpindex(const char *, regex_t *, regmatch_t *);
209 static void do_regexp(const char *, regex_t *, const char *, regmatch_t *);
210 static void add_sub(size_t, const char *, regex_t *, regmatch_t *);
211 static void add_replace(const char *, regex_t *, const char *, regmatch_t *);
212 #define addconstantstring(s) addchars((s), sizeof(s)-1)
213 
214 static void
215 addchars(const char *c, size_t n)
216 {
217 	if (n == 0)
218 		return;
219 	while (current + n > bufsize) {
220 		if (bufsize == 0)
221 			bufsize = 1024;
222 		else
223 			bufsize *= 2;
224 		buffer = xrealloc(buffer, bufsize, NULL);
225 	}
226 	memcpy(buffer+current, c, n);
227 	current += n;
228 }
229 
230 static void
231 addchar(int c)
232 {
233 	if (current +1 > bufsize) {
234 		if (bufsize == 0)
235 			bufsize = 1024;
236 		else
237 			bufsize *= 2;
238 		buffer = xrealloc(buffer, bufsize, NULL);
239 	}
240 	buffer[current++] = c;
241 }
242 
243 static char *
244 getstring()
245 {
246 	addchar('\0');
247 	current = 0;
248 	return buffer;
249 }
250 
251 
252 static void
253 exit_regerror(int er, regex_t *re)
254 {
255 	size_t 	errlen;
256 	char 	*errbuf;
257 
258 	errlen = regerror(er, re, NULL, 0);
259 	errbuf = xalloc(errlen,
260 	    "malloc in regerror: %lu", (unsigned long)errlen);
261 	regerror(er, re, errbuf, errlen);
262 	m4errx(1, "regular expression error: %s.", errbuf);
263 }
264 
265 static void
266 add_sub(size_t n, const char *string, regex_t *re, regmatch_t *pm)
267 {
268 	if (n > re->re_nsub)
269 		warnx("No subexpression %zu", n);
270 	/* Subexpressions that did not match are
271 	 * not an error.  */
272 	else if (pm[n].rm_so != -1 &&
273 	    pm[n].rm_eo != -1) {
274 		addchars(string + pm[n].rm_so,
275 			pm[n].rm_eo - pm[n].rm_so);
276 	}
277 }
278 
279 /* Add replacement string to the output buffer, recognizing special
280  * constructs and replacing them with substrings of the original string.
281  */
282 static void
283 add_replace(const char *string, regex_t *re, const char *replace, regmatch_t *pm)
284 {
285 	const char *p;
286 
287 	for (p = replace; *p != '\0'; p++) {
288 		if (*p == '&' && !mimic_gnu) {
289 			add_sub(0, string, re, pm);
290 			continue;
291 		}
292 		if (*p == '\\') {
293 			if (p[1] == '\\') {
294 				addchar(p[1]);
295 				p++;
296 				continue;
297 			}
298 			if (p[1] == '&') {
299 				if (mimic_gnu)
300 					add_sub(0, string, re, pm);
301 				else
302 					addchar(p[1]);
303 				p++;
304 				continue;
305 			}
306 			if (isdigit((unsigned char)p[1])) {
307 				add_sub(*(++p) - '0', string, re, pm);
308 				continue;
309 			}
310 		}
311 	    	addchar(*p);
312 	}
313 }
314 
315 static void
316 do_subst(const char *string, regex_t *re, const char *replace, regmatch_t *pm)
317 {
318 	int error;
319 	int flags = 0;
320 	const char *last_match = NULL;
321 
322 	while ((error = regexec(re, string, re->re_nsub+1, pm, flags)) == 0) {
323 		if (pm[0].rm_eo != 0) {
324 			if (string[pm[0].rm_eo-1] == '\n')
325 				flags = 0;
326 			else
327 				flags = REG_NOTBOL;
328 		}
329 
330 		/* NULL length matches are special... We use the `vi-mode'
331 		 * rule: don't allow a NULL-match at the last match
332 		 * position.
333 		 */
334 		if (pm[0].rm_so == pm[0].rm_eo &&
335 		    string + pm[0].rm_so == last_match) {
336 			if (*string == '\0')
337 				return;
338 			addchar(*string);
339 			if (*string++ == '\n')
340 				flags = 0;
341 			else
342 				flags = REG_NOTBOL;
343 			continue;
344 		}
345 		last_match = string + pm[0].rm_so;
346 		addchars(string, pm[0].rm_so);
347 		add_replace(string, re, replace, pm);
348 		string += pm[0].rm_eo;
349 	}
350 	if (error != REG_NOMATCH)
351 		exit_regerror(error, re);
352 	pbstr(string);
353 }
354 
355 static void
356 do_regexp(const char *string, regex_t *re, const char *replace, regmatch_t *pm)
357 {
358 	int error;
359 
360 	switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
361 	case 0:
362 		add_replace(string, re, replace, pm);
363 		pbstr(getstring());
364 		break;
365 	case REG_NOMATCH:
366 		break;
367 	default:
368 		exit_regerror(error, re);
369 	}
370 }
371 
372 static void
373 do_regexpindex(const char *string, regex_t *re, regmatch_t *pm)
374 {
375 	int error;
376 
377 	switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
378 	case 0:
379 		pbunsigned(pm[0].rm_so);
380 		break;
381 	case REG_NOMATCH:
382 		pbnum(-1);
383 		break;
384 	default:
385 		exit_regerror(error, re);
386 	}
387 }
388 
389 /* In Gnu m4 mode, parentheses for backmatch don't work like POSIX 1003.2
390  * says. So we twiddle with the regexp before passing it to regcomp.
391  */
392 static char *
393 twiddle(const char *p)
394 {
395 	/* + at start of regexp is a normal character for Gnu m4 */
396 	if (*p == '^') {
397 		addchar(*p);
398 		p++;
399 	}
400 	if (*p == '+') {
401 		addchar('\\');
402 	}
403 	/* This could use strcspn for speed... */
404 	while (*p != '\0') {
405 		if (*p == '\\') {
406 			switch(p[1]) {
407 			case '(':
408 			case ')':
409 			case '|':
410 				addchar(p[1]);
411 				break;
412 			case 'w':
413 				addconstantstring("[_a-zA-Z0-9]");
414 				break;
415 			case 'W':
416 				addconstantstring("[^_a-zA-Z0-9]");
417 				break;
418 			case '<':
419 				addconstantstring("[[:<:]]");
420 				break;
421 			case '>':
422 				addconstantstring("[[:>:]]");
423 				break;
424 			default:
425 				addchars(p, 2);
426 				break;
427 			}
428 			p+=2;
429 			continue;
430 		}
431 		if (*p == '(' || *p == ')' || *p == '|')
432 			addchar('\\');
433 
434 		addchar(*p);
435 		p++;
436 	}
437 	return getstring();
438 }
439 
440 /* patsubst(string, regexp, opt replacement) */
441 /* argv[2]: string
442  * argv[3]: regexp
443  * argv[4]: opt rep
444  */
445 void
446 dopatsubst(const char *argv[], int argc)
447 {
448 	if (argc <= 3) {
449 		warnx("Too few arguments to patsubst");
450 		return;
451 	}
452 	/* special case: empty regexp */
453 	if (argv[3][0] == '\0') {
454 		const char *s;
455 		size_t len;
456 		if (argv[4] && argc > 4)
457 			len = strlen(argv[4]);
458 		else
459 			len = 0;
460 		for (s = argv[2]; *s != '\0'; s++) {
461 			addchars(argv[4], len);
462 			addchar(*s);
463 		}
464 	} else {
465 		int error;
466 		regex_t re;
467 		regmatch_t *pmatch;
468 		int mode = REG_EXTENDED;
469 		size_t l = strlen(argv[3]);
470 
471 		if (!mimic_gnu ||
472 		    (argv[3][0] == '^') ||
473 		    (l > 0 && argv[3][l-1] == '$'))
474 			mode |= REG_NEWLINE;
475 
476 		error = regcomp(&re, mimic_gnu ? twiddle(argv[3]) : argv[3],
477 		    mode);
478 		if (error != 0)
479 			exit_regerror(error, &re);
480 
481 		pmatch = xalloc(sizeof(regmatch_t) * (re.re_nsub+1), NULL);
482 		do_subst(argv[2], &re,
483 		    argc > 4 && argv[4] != NULL ? argv[4] : "", pmatch);
484 		free(pmatch);
485 		regfree(&re);
486 	}
487 	pbstr(getstring());
488 }
489 
490 void
491 doregexp(const char *argv[], int argc)
492 {
493 	int error;
494 	regex_t re;
495 	regmatch_t *pmatch;
496 
497 	if (argc <= 3) {
498 		warnx("Too few arguments to regexp");
499 		return;
500 	}
501 	error = regcomp(&re, mimic_gnu ? twiddle(argv[3]) : argv[3],
502 	    REG_EXTENDED);
503 	if (error != 0)
504 		exit_regerror(error, &re);
505 
506 	pmatch = xalloc(sizeof(regmatch_t) * (re.re_nsub+1), NULL);
507 	if (argv[4] == NULL || argc == 4)
508 		do_regexpindex(argv[2], &re, pmatch);
509 	else
510 		do_regexp(argv[2], &re, argv[4], pmatch);
511 	free(pmatch);
512 	regfree(&re);
513 }
514 
515 void
516 doformat(const char *argv[], int argc)
517 {
518 	const char *format = argv[2];
519 	int pos = 3;
520 	int left_padded;
521 	long width;
522 	size_t l;
523 	const char *thisarg;
524 	char temp[2];
525 	size_t extra;
526 
527 	while (*format != 0) {
528 		if (*format != '%') {
529 			addchar(*format++);
530 			continue;
531 		}
532 
533 		format++;
534 		if (*format == '%') {
535 			addchar(*format++);
536 			continue;
537 		}
538 		if (*format == 0) {
539 			addchar('%');
540 			break;
541 		}
542 
543 		if (*format == '*') {
544 			format++;
545 			if (pos >= argc)
546 				m4errx(1,
547 				    "Format with too many format specifiers.");
548 			width = strtol(argv[pos++], NULL, 10);
549 		} else {
550 			char *eformat;
551 			width = strtol(format, &eformat, 10);
552 			format = eformat;
553 		}
554 		if (width < 0) {
555 			left_padded = 1;
556 			width = -width;
557 		} else {
558 			left_padded = 0;
559 		}
560 		if (*format == '.') {
561 			format++;
562 			if (*format == '*') {
563 				format++;
564 				if (pos >= argc)
565 					m4errx(1,
566 					    "Format with too many format specifiers.");
567 				extra = strtol(argv[pos++], NULL, 10);
568 			} else {
569 				char *eformat;
570 				extra = strtol(format, &eformat, 10);
571 				format = eformat;
572 			}
573 		} else {
574 			extra = SIZE_T_MAX;
575 		}
576 		if (pos >= argc)
577 			m4errx(1, "Format with too many format specifiers.");
578 		switch(*format) {
579 		case 's':
580 			thisarg = argv[pos++];
581 			break;
582 		case 'c':
583 			temp[0] = strtoul(argv[pos++], NULL, 10);
584 			temp[1] = 0;
585 			thisarg = temp;
586 			break;
587 		default:
588 			m4errx(1, "Unsupported format specification: %s.",
589 			    argv[2]);
590 		}
591 		format++;
592 		l = strlen(thisarg);
593 		if (l > extra)
594 			l = extra;
595 		if (!left_padded) {
596 			while (l < (size_t)width--)
597 				addchar(' ');
598 		}
599 		addchars(thisarg, l);
600 		if (left_padded) {
601 			while (l < (size_t)width--)
602 				addchar(' ');
603 		}
604 	}
605 	pbstr(getstring());
606 }
607 
608 void
609 doesyscmd(const char *cmd)
610 {
611 	int p[2];
612 	pid_t pid, cpid;
613 	const char *argv[4];
614 	int cc;
615 	int status;
616 
617 	/* Follow gnu m4 documentation: first flush buffers. */
618 	fflush(NULL);
619 
620 	argv[0] = "sh";
621 	argv[1] = "-c";
622 	argv[2] = cmd;
623 	argv[3] = NULL;
624 
625 	/* Just set up standard output, share stderr and stdin with m4 */
626 	if (pipe(p) == -1)
627 		err(1, "bad pipe");
628 	switch(cpid = fork()) {
629 	case -1:
630 		err(1, "bad fork");
631 		/* NOTREACHED */
632 	case 0:
633 		(void) close(p[0]);
634 		(void) dup2(p[1], 1);
635 		(void) close(p[1]);
636 		execv(_PATH_BSHELL, __UNCONST(argv));
637 		exit(1);
638 	default:
639 		/* Read result in two stages, since m4's buffer is
640 		 * pushback-only. */
641 		(void) close(p[1]);
642 		do {
643 			char result[BUFSIZE];
644 			cc = read(p[0], result, sizeof result);
645 			if (cc > 0)
646 				addchars(result, cc);
647 		} while (cc > 0 || (cc == -1 && errno == EINTR));
648 
649 		(void) close(p[0]);
650 		while ((pid = wait(&status)) != cpid && pid >= 0)
651 			continue;
652 		pbstr(getstring());
653 	}
654 }
655 
656 void
657 getdivfile(const char *name)
658 {
659 	FILE *f;
660 	int c;
661 
662 	f = fopen(name, "r");
663 	if (!f)
664 		return;
665 
666 	while ((c = getc(f))!= EOF)
667 		putc(c, active);
668 	(void) fclose(f);
669 }
670