xref: /openbsd/usr.bin/m4/gnum4.c (revision 4cfece93) 
1 /* $OpenBSD: gnum4.c,v 1.52 2017/08/21 21:41:13 deraadt Exp $ */
2 
3 /*
4  * Copyright (c) 1999 Marc Espie
5  *
6  * Redistribution and use in source and binary forms, with or without
7  * modification, are permitted provided that the following conditions
8  * are met:
9  * 1. Redistributions of source code must retain the above copyright
10  *    notice, this list of conditions and the following disclaimer.
11  * 2. Redistributions in binary form must reproduce the above copyright
12  *    notice, this list of conditions and the following disclaimer in the
13  *    documentation and/or other materials provided with the distribution.
14  *
15  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
16  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
17  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
18  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
19  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
20  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
21  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
22  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
23  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
24  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
25  * SUCH DAMAGE.
26  */
27 
28 /*
29  * functions needed to support gnu-m4 extensions, including a fake freezing
30  */
31 
32 #include <sys/types.h>
33 #include <sys/wait.h>
34 #include <ctype.h>
35 #include <err.h>
36 #include <paths.h>
37 #include <regex.h>
38 #include <stdarg.h>
39 #include <stddef.h>
40 #include <stdlib.h>
41 #include <stdint.h>
42 #include <stdio.h>
43 #include <string.h>
44 #include <errno.h>
45 #include <unistd.h>
46 #include <limits.h>
47 #include "mdef.h"
48 #include "stdd.h"
49 #include "extern.h"
50 
51 
52 int mimic_gnu = 0;
53 
54 /*
55  * Support for include path search
56  * First search in the current directory.
57  * If not found, and the path is not absolute, include path kicks in.
58  * First, -I options, in the order found on the command line.
59  * Then M4PATH env variable
60  */
61 
62 struct path_entry {
63 	char *name;
64 	struct path_entry *next;
65 } *first, *last;
66 
67 static struct path_entry *new_path_entry(const char *);
68 static void ensure_m4path(void);
69 static struct input_file *dopath(struct input_file *, const char *);
70 
71 static struct path_entry *
72 new_path_entry(const char *dirname)
73 {
74 	struct path_entry *n;
75 
76 	n = malloc(sizeof(struct path_entry));
77 	if (!n)
78 		errx(1, "out of memory");
79 	n->name = xstrdup(dirname);
80 	n->next = 0;
81 	return n;
82 }
83 
84 void
85 addtoincludepath(const char *dirname)
86 {
87 	struct path_entry *n;
88 
89 	n = new_path_entry(dirname);
90 
91 	if (last) {
92 		last->next = n;
93 		last = n;
94 	}
95 	else
96 		last = first = n;
97 }
98 
99 static void
100 ensure_m4path()
101 {
102 	static int envpathdone = 0;
103 	char *envpath;
104 	char *sweep;
105 	char *path;
106 
107 	if (envpathdone)
108 		return;
109 	envpathdone = TRUE;
110 	envpath = getenv("M4PATH");
111 	if (!envpath)
112 		return;
113 	/* for portability: getenv result is read-only */
114 	envpath = xstrdup(envpath);
115 	for (sweep = envpath;
116 	    (path = strsep(&sweep, ":")) != NULL;)
117 	    addtoincludepath(path);
118 	free(envpath);
119 }
120 
121 static
122 struct input_file *
123 dopath(struct input_file *i, const char *filename)
124 {
125 	char path[PATH_MAX];
126 	struct path_entry *pe;
127 	FILE *f;
128 
129 	for (pe = first; pe; pe = pe->next) {
130 		snprintf(path, sizeof(path), "%s/%s", pe->name, filename);
131 		if ((f = fopen(path, "r")) != 0) {
132 			set_input(i, f, path);
133 			return i;
134 		}
135 	}
136 	return NULL;
137 }
138 
139 struct input_file *
140 fopen_trypath(struct input_file *i, const char *filename)
141 {
142 	FILE *f;
143 
144 	f = fopen(filename, "r");
145 	if (f != NULL) {
146 		set_input(i, f, filename);
147 		return i;
148 	}
149 	if (filename[0] == '/')
150 		return NULL;
151 
152 	ensure_m4path();
153 
154 	return dopath(i, filename);
155 }
156 
157 void
158 doindir(const char *argv[], int argc)
159 {
160 	ndptr n;
161 	struct macro_definition *p;
162 
163 	n = lookup(argv[2]);
164 	if (n == NULL || (p = macro_getdef(n)) == NULL)
165 		m4errx(1, "indir: undefined macro %s.", argv[2]);
166 	argv[1] = p->defn;
167 
168 	eval(argv+1, argc-1, p->type, is_traced(n));
169 }
170 
171 void
172 dobuiltin(const char *argv[], int argc)
173 {
174 	ndptr p;
175 
176 	argv[1] = NULL;
177 	p = macro_getbuiltin(argv[2]);
178 	if (p != NULL)
179 		eval(argv+1, argc-1, macro_builtin_type(p), is_traced(p));
180 	else
181 		m4errx(1, "unknown builtin %s.", argv[2]);
182 }
183 
184 
185 /* We need some temporary buffer space, as pb pushes BACK and substitution
186  * proceeds forward... */
187 static char *buffer;
188 static size_t bufsize = 0;
189 static size_t current = 0;
190 
191 static void addchars(const char *, size_t);
192 static void addchar(int);
193 static char *twiddle(const char *);
194 static char *getstring(void);
195 static void exit_regerror(int, regex_t *, const char *);
196 static void do_subst(const char *, regex_t *, const char *, const char *,
197     regmatch_t *);
198 static void do_regexpindex(const char *, regex_t *, const char *, regmatch_t *);
199 static void do_regexp(const char *, regex_t *, const char *, const char *,
200     regmatch_t *);
201 static void add_sub(int, const char *, regex_t *, regmatch_t *);
202 static void add_replace(const char *, regex_t *, const char *, regmatch_t *);
203 #define addconstantstring(s) addchars((s), sizeof(s)-1)
204 
205 static void
206 addchars(const char *c, size_t n)
207 {
208 	if (n == 0)
209 		return;
210 	while (current + n > bufsize) {
211 		if (bufsize == 0)
212 			bufsize = 1024;
213 		else if (bufsize <= SIZE_MAX/2) {
214 			bufsize *= 2;
215 		} else {
216 			errx(1, "size overflow");
217 		}
218 		buffer = xrealloc(buffer, bufsize, NULL);
219 	}
220 	memcpy(buffer+current, c, n);
221 	current += n;
222 }
223 
224 static void
225 addchar(int c)
226 {
227 	if (current +1 > bufsize) {
228 		if (bufsize == 0)
229 			bufsize = 1024;
230 		else
231 			bufsize *= 2;
232 		buffer = xrealloc(buffer, bufsize, NULL);
233 	}
234 	buffer[current++] = c;
235 }
236 
237 static char *
238 getstring(void)
239 {
240 	addchar('\0');
241 	current = 0;
242 	return buffer;
243 }
244 
245 
246 static void
247 exit_regerror(int er, regex_t *re, const char *source)
248 {
249 	size_t	errlen;
250 	char	*errbuf;
251 
252 	errlen = regerror(er, re, NULL, 0);
253 	errbuf = xalloc(errlen,
254 	    "malloc in regerror: %lu", (unsigned long)errlen);
255 	regerror(er, re, errbuf, errlen);
256 	m4errx(1, "regular expression error in %s: %s.", source, errbuf);
257 }
258 
259 /* warnx() plus check to see if we need to change exit code or exit.
260  * -E flag functionality.
261  */
262 void
263 m4_warnx(const char *fmt, ...)
264 {
265 	va_list ap;
266 
267 	va_start(ap, fmt);
268 	vwarnx(fmt, ap);
269 	va_end(ap);
270 
271 	if (fatal_warns)
272 		exit(1);
273 	if (error_warns)
274 		exit_code = 1;
275 }
276 
277 static void
278 add_sub(int n, const char *string, regex_t *re, regmatch_t *pm)
279 {
280 	if (n > re->re_nsub)
281 		m4_warnx("No subexpression %d", n);
282 	/* Subexpressions that did not match are
283 	 * not an error.  */
284 	else if (pm[n].rm_so != -1 &&
285 	    pm[n].rm_eo != -1) {
286 		addchars(string + pm[n].rm_so,
287 			pm[n].rm_eo - pm[n].rm_so);
288 	}
289 }
290 
291 /* Add replacement string to the output buffer, recognizing special
292  * constructs and replacing them with substrings of the original string.
293  */
294 static void
295 add_replace(const char *string, regex_t *re, const char *replace, regmatch_t *pm)
296 {
297 	const char *p;
298 
299 	for (p = replace; *p != '\0'; p++) {
300 		if (*p == '&' && !mimic_gnu) {
301 			add_sub(0, string, re, pm);
302 			continue;
303 		}
304 		if (*p == '\\') {
305 			if (p[1] == '\\') {
306 				addchar(p[1]);
307 				p++;
308 				continue;
309 			}
310 			if (p[1] == '&') {
311 				if (mimic_gnu)
312 					add_sub(0, string, re, pm);
313 				else
314 					addchar(p[1]);
315 				p++;
316 				continue;
317 			}
318 			if (isdigit((unsigned char)p[1])) {
319 				add_sub(*(++p) - '0', string, re, pm);
320 				continue;
321 			}
322 		}
323 		addchar(*p);
324 	}
325 }
326 
327 static void
328 do_subst(const char *string, regex_t *re, const char *source,
329     const char *replace, regmatch_t *pm)
330 {
331 	int error;
332 	int flags = 0;
333 	const char *last_match = NULL;
334 
335 	while ((error = regexec(re, string, re->re_nsub+1, pm, flags)) == 0) {
336 		if (pm[0].rm_eo != 0) {
337 			if (string[pm[0].rm_eo-1] == '\n')
338 				flags = 0;
339 			else
340 				flags = REG_NOTBOL;
341 		}
342 
343 		/* NULL length matches are special... We use the `vi-mode'
344 		 * rule: don't allow a NULL-match at the last match
345 		 * position.
346 		 */
347 		if (pm[0].rm_so == pm[0].rm_eo &&
348 		    string + pm[0].rm_so == last_match) {
349 			if (*string == '\0')
350 				return;
351 			addchar(*string);
352 			if (*string++ == '\n')
353 				flags = 0;
354 			else
355 				flags = REG_NOTBOL;
356 			continue;
357 		}
358 		last_match = string + pm[0].rm_so;
359 		addchars(string, pm[0].rm_so);
360 		add_replace(string, re, replace, pm);
361 		string += pm[0].rm_eo;
362 	}
363 	if (error != REG_NOMATCH)
364 		exit_regerror(error, re, source);
365 	pbstr(string);
366 }
367 
368 static void
369 do_regexp(const char *string, regex_t *re, const char *source,
370     const char *replace, regmatch_t *pm)
371 {
372 	int error;
373 
374 	switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
375 	case 0:
376 		add_replace(string, re, replace, pm);
377 		pbstr(getstring());
378 		break;
379 	case REG_NOMATCH:
380 		break;
381 	default:
382 		exit_regerror(error, re, source);
383 	}
384 }
385 
386 static void
387 do_regexpindex(const char *string, regex_t *re, const char *source,
388     regmatch_t *pm)
389 {
390 	int error;
391 
392 	switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
393 	case 0:
394 		pbunsigned(pm[0].rm_so);
395 		break;
396 	case REG_NOMATCH:
397 		pbnum(-1);
398 		break;
399 	default:
400 		exit_regerror(error, re, source);
401 	}
402 }
403 
404 /* In Gnu m4 mode, parentheses for backmatch don't work like POSIX 1003.2
405  * says. So we twiddle with the regexp before passing it to regcomp.
406  */
407 static char *
408 twiddle(const char *p)
409 {
410 	/* + at start of regexp is a normal character for Gnu m4 */
411 	if (*p == '^') {
412 		addchar(*p);
413 		p++;
414 	}
415 	if (*p == '+') {
416 		addchar('\\');
417 	}
418 	/* This could use strcspn for speed... */
419 	while (*p != '\0') {
420 		if (*p == '\\') {
421 			switch(p[1]) {
422 			case '(':
423 			case ')':
424 			case '|':
425 				addchar(p[1]);
426 				break;
427 			case 'w':
428 				addconstantstring("[_a-zA-Z0-9]");
429 				break;
430 			case 'W':
431 				addconstantstring("[^_a-zA-Z0-9]");
432 				break;
433 			case '<':
434 				addconstantstring("[[:<:]]");
435 				break;
436 			case '>':
437 				addconstantstring("[[:>:]]");
438 				break;
439 			default:
440 				addchars(p, 2);
441 				break;
442 			}
443 			p+=2;
444 			continue;
445 		}
446 		if (*p == '(' || *p == ')' || *p == '|')
447 			addchar('\\');
448 
449 		addchar(*p);
450 		p++;
451 	}
452 	return getstring();
453 }
454 
455 /* patsubst(string, regexp, opt replacement) */
456 /* argv[2]: string
457  * argv[3]: regexp
458  * argv[4]: opt rep
459  */
460 void
461 dopatsubst(const char *argv[], int argc)
462 {
463 	if (argc <= 3) {
464 		m4_warnx("Too few arguments to patsubst");
465 		return;
466 	}
467 	/* special case: empty regexp */
468 	if (argv[3][0] == '\0') {
469 		const char *s;
470 		size_t len;
471 		if (argc > 4 && argv[4])
472 			len = strlen(argv[4]);
473 		else
474 			len = 0;
475 		for (s = argv[2]; *s != '\0'; s++) {
476 			addchars(argv[4], len);
477 			addchar(*s);
478 		}
479 	} else {
480 		int error;
481 		regex_t re;
482 		regmatch_t *pmatch;
483 		int mode = REG_EXTENDED;
484 		const char *source;
485 		size_t l = strlen(argv[3]);
486 
487 		if (!mimic_gnu ||
488 		    (argv[3][0] == '^') ||
489 		    (l > 0 && argv[3][l-1] == '$'))
490 			mode |= REG_NEWLINE;
491 
492 		source = mimic_gnu ? twiddle(argv[3]) : argv[3];
493 		error = regcomp(&re, source, mode);
494 		if (error != 0)
495 			exit_regerror(error, &re, source);
496 
497 		pmatch = xreallocarray(NULL, re.re_nsub+1, sizeof(regmatch_t),
498 		    NULL);
499 		do_subst(argv[2], &re, source,
500 		    argc > 4 && argv[4] != NULL ? argv[4] : "", pmatch);
501 		free(pmatch);
502 		regfree(&re);
503 	}
504 	pbstr(getstring());
505 }
506 
507 void
508 doregexp(const char *argv[], int argc)
509 {
510 	int error;
511 	regex_t re;
512 	regmatch_t *pmatch;
513 	const char *source;
514 
515 	if (argc <= 3) {
516 		m4_warnx("Too few arguments to regexp");
517 		return;
518 	}
519 	/* special gnu case */
520 	if (argv[3][0] == '\0' && mimic_gnu) {
521 		if (argc == 4 || argv[4] == NULL)
522 			return;
523 		else
524 			pbstr(argv[4]);
525 	}
526 	source = mimic_gnu ? twiddle(argv[3]) : argv[3];
527 	error = regcomp(&re, source, REG_EXTENDED|REG_NEWLINE);
528 	if (error != 0)
529 		exit_regerror(error, &re, source);
530 
531 	pmatch = xreallocarray(NULL, re.re_nsub+1, sizeof(regmatch_t), NULL);
532 	if (argc == 4 || argv[4] == NULL)
533 		do_regexpindex(argv[2], &re, source, pmatch);
534 	else
535 		do_regexp(argv[2], &re, source, argv[4], pmatch);
536 	free(pmatch);
537 	regfree(&re);
538 }
539 
540 void
541 doformat(const char *argv[], int argc)
542 {
543 	const char *format = argv[2];
544 	int pos = 3;
545 	int left_padded;
546 	long width;
547 	size_t l;
548 	const char *thisarg;
549 	char temp[2];
550 	long extra;
551 
552 	while (*format != 0) {
553 		if (*format != '%') {
554 			addchar(*format++);
555 			continue;
556 		}
557 
558 		format++;
559 		if (*format == '%') {
560 			addchar(*format++);
561 			continue;
562 		}
563 		if (*format == 0) {
564 			addchar('%');
565 			break;
566 		}
567 
568 		if (*format == '*') {
569 			format++;
570 			if (pos >= argc)
571 				m4errx(1,
572 				    "Format with too many format specifiers.");
573 			width = strtol(argv[pos++], NULL, 10);
574 		} else {
575 			width = strtol(format, (char **)&format, 10);
576 		}
577 		if (width < 0) {
578 			left_padded = 1;
579 			width = -width;
580 		} else {
581 			left_padded = 0;
582 		}
583 		if (*format == '.') {
584 			format++;
585 			if (*format == '*') {
586 				format++;
587 				if (pos >= argc)
588 					m4errx(1,
589 					    "Format with too many format specifiers.");
590 				extra = strtol(argv[pos++], NULL, 10);
591 			} else {
592 				extra = strtol(format, (char **)&format, 10);
593 			}
594 		} else {
595 			extra = LONG_MAX;
596 		}
597 		if (pos >= argc)
598 			m4errx(1, "Format with too many format specifiers.");
599 		switch(*format) {
600 		case 's':
601 			thisarg = argv[pos++];
602 			break;
603 		case 'c':
604 			temp[0] = strtoul(argv[pos++], NULL, 10);
605 			temp[1] = 0;
606 			thisarg = temp;
607 			break;
608 		default:
609 			m4errx(1, "Unsupported format specification: %s.",
610 			    argv[2]);
611 		}
612 		format++;
613 		l = strlen(thisarg);
614 		if (l > extra)
615 			l = extra;
616 		if (!left_padded) {
617 			while (l < width--)
618 				addchar(' ');
619 		}
620 		addchars(thisarg, l);
621 		if (left_padded) {
622 			while (l < width--)
623 				addchar(' ');
624 		}
625 	}
626 	pbstr(getstring());
627 }
628 
629 void
630 doesyscmd(const char *cmd)
631 {
632 	int p[2];
633 	pid_t cpid;
634 	char *argv[4];
635 	int cc;
636 	int status;
637 
638 	/* Follow gnu m4 documentation: first flush buffers. */
639 	fflush(NULL);
640 
641 	argv[0] = "sh";
642 	argv[1] = "-c";
643 	argv[2] = (char *)cmd;
644 	argv[3] = NULL;
645 
646 	/* Just set up standard output, share stderr and stdin with m4 */
647 	if (pipe(p) == -1)
648 		err(1, "bad pipe");
649 	switch(cpid = fork()) {
650 	case -1:
651 		err(1, "bad fork");
652 		/* NOTREACHED */
653 	case 0:
654 		(void) close(p[0]);
655 		(void) dup2(p[1], 1);
656 		(void) close(p[1]);
657 		execv(_PATH_BSHELL, argv);
658 		exit(1);
659 	default:
660 		/* Read result in two stages, since m4's buffer is
661 		 * pushback-only. */
662 		(void) close(p[1]);
663 		do {
664 			char result[BUFSIZE];
665 			cc = read(p[0], result, sizeof result);
666 			if (cc > 0)
667 				addchars(result, cc);
668 		} while (cc > 0 || (cc == -1 && errno == EINTR));
669 
670 		(void) close(p[0]);
671 		while (waitpid(cpid, &status, 0) == -1) {
672 			if (errno != EINTR)
673 				break;
674 		}
675 		pbstr(getstring());
676 	}
677 }
678 
679 void
680 getdivfile(const char *name)
681 {
682 	FILE *f;
683 	int c;
684 
685 	f = fopen(name, "r");
686 	if (!f)
687 		return;
688 
689 	while ((c = getc(f))!= EOF)
690 		putc(c, active);
691 	(void) fclose(f);
692 }
693