xref: /freebsd/usr.bin/m4/gnum4.c (revision 74d9553e43cfafc29448d0bb836916aa21dea0de) 
1 /* $OpenBSD: gnum4.c,v 1.50 2015/04/29 00:13:26 millert 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 #include <sys/cdefs.h>
28 __FBSDID("$FreeBSD$");
29 
30 /*
31  * functions needed to support gnu-m4 extensions, including a fake freezing
32  */
33 
34 #include <sys/types.h>
35 #include <sys/wait.h>
36 #include <ctype.h>
37 #include <err.h>
38 #include <paths.h>
39 #include <regex.h>
40 #include <stddef.h>
41 #include <stdlib.h>
42 #include <stdint.h>
43 #include <stdio.h>
44 #include <string.h>
45 #include <errno.h>
46 #include <unistd.h>
47 #include <limits.h>
48 #include "mdef.h"
49 #include "stdd.h"
50 #include "extern.h"
51 
52 
53 int mimic_gnu = 0;
54 
55 /*
56  * Support for include path search
57  * First search in the current directory.
58  * If not found, and the path is not absolute, include path kicks in.
59  * First, -I options, in the order found on the command line.
60  * Then M4PATH env variable
61  */
62 
63 static struct path_entry {
64 	char *name;
65 	struct path_entry *next;
66 } *first, *last;
67 
68 static struct path_entry *new_path_entry(const char *);
69 static void ensure_m4path(void);
70 static struct input_file *dopath(struct input_file *, const char *);
71 
72 static struct path_entry *
73 new_path_entry(const char *dirname)
74 {
75 	struct path_entry *n;
76 
77 	n = malloc(sizeof(struct path_entry));
78 	if (!n)
79 		errx(1, "out of memory");
80 	n->name = xstrdup(dirname);
81 	n->next = 0;
82 	return n;
83 }
84 
85 void
86 addtoincludepath(const char *dirname)
87 {
88 	struct path_entry *n;
89 
90 	n = new_path_entry(dirname);
91 
92 	if (last) {
93 		last->next = n;
94 		last = n;
95 	}
96 	else
97 		last = first = n;
98 }
99 
100 static void
101 ensure_m4path(void)
102 {
103 	static int envpathdone = 0;
104 	char *envpath;
105 	char *sweep;
106 	char *path;
107 
108 	if (envpathdone)
109 		return;
110 	envpathdone = TRUE;
111 	envpath = getenv("M4PATH");
112 	if (!envpath)
113 		return;
114 	/* for portability: getenv result is read-only */
115 	envpath = xstrdup(envpath);
116 	for (sweep = envpath;
117 	    (path = strsep(&sweep, ":")) != NULL;)
118 	    addtoincludepath(path);
119 	free(envpath);
120 }
121 
122 static
123 struct input_file *
124 dopath(struct input_file *i, const char *filename)
125 {
126 	char path[PATH_MAX];
127 	struct path_entry *pe;
128 	FILE *f;
129 
130 	for (pe = first; pe; pe = pe->next) {
131 		snprintf(path, sizeof(path), "%s/%s", pe->name, filename);
132 		if ((f = fopen(path, "r")) != NULL) {
133 			set_input(i, f, path);
134 			return i;
135 		}
136 	}
137 	return NULL;
138 }
139 
140 struct input_file *
141 fopen_trypath(struct input_file *i, const char *filename)
142 {
143 	FILE *f;
144 
145 	f = fopen(filename, "r");
146 	if (f != NULL) {
147 		set_input(i, f, filename);
148 		return i;
149 	}
150 	if (filename[0] == '/')
151 		return NULL;
152 
153 	ensure_m4path();
154 
155 	return dopath(i, filename);
156 }
157 
158 void
159 doindir(const char *argv[], int argc)
160 {
161 	ndptr n;
162 	struct macro_definition *p = NULL;
163 
164 	n = lookup(argv[2]);
165 	if (n == NULL || (p = macro_getdef(n)) == NULL)
166 		m4errx(1, "indir: undefined macro %s.", argv[2]);
167 	argv[1] = p->defn;
168 
169 	eval(argv+1, argc-1, p->type, is_traced(n));
170 }
171 
172 void
173 dobuiltin(const char *argv[], int argc)
174 {
175 	ndptr p;
176 
177 	argv[1] = NULL;
178 	p = macro_getbuiltin(argv[2]);
179 	if (p != NULL)
180 		eval(argv+1, argc-1, macro_builtin_type(p), is_traced(p));
181 	else
182 		m4errx(1, "unknown builtin %s.", argv[2]);
183 }
184 
185 
186 /* We need some temporary buffer space, as pb pushes BACK and substitution
187  * proceeds forward... */
188 static char *buffer;
189 static size_t bufsize = 0;
190 static size_t current = 0;
191 
192 static void addchars(const char *, size_t);
193 static void addchar(int);
194 static char *twiddle(const char *);
195 static char *getstring(void);
196 static void exit_regerror(int, regex_t *, const char *);
197 static void do_subst(const char *, regex_t *, const char *, const char *,
198     regmatch_t *);
199 static void do_regexpindex(const char *, regex_t *, const char *, regmatch_t *);
200 static void do_regexp(const char *, regex_t *, const char *, const char *,
201     regmatch_t *);
202 static void add_sub(int, const char *, regex_t *, regmatch_t *);
203 static void add_replace(const char *, regex_t *, const char *, regmatch_t *);
204 #define addconstantstring(s) addchars((s), sizeof(s)-1)
205 
206 static void
207 addchars(const char *c, size_t n)
208 {
209 	if (n == 0)
210 		return;
211 	while (current + n > bufsize) {
212 		if (bufsize == 0)
213 			bufsize = 1024;
214 		else if (bufsize <= SIZE_MAX/2) {
215 			bufsize *= 2;
216 		} else {
217 			errx(1, "size overflow");
218 		}
219 		buffer = xrealloc(buffer, bufsize, NULL);
220 	}
221 	memcpy(buffer+current, c, n);
222 	current += n;
223 }
224 
225 static void
226 addchar(int c)
227 {
228 	if (current +1 > bufsize) {
229 		if (bufsize == 0)
230 			bufsize = 1024;
231 		else
232 			bufsize *= 2;
233 		buffer = xrealloc(buffer, bufsize, NULL);
234 	}
235 	buffer[current++] = c;
236 }
237 
238 static char *
239 getstring(void)
240 {
241 	addchar('\0');
242 	current = 0;
243 	return buffer;
244 }
245 
246 
247 static void
248 exit_regerror(int er, regex_t *re, const char *source)
249 {
250 	size_t	errlen;
251 	char	*errbuf;
252 
253 	errlen = regerror(er, re, NULL, 0);
254 	errbuf = xalloc(errlen,
255 	    "malloc in regerror: %lu", (unsigned long)errlen);
256 	regerror(er, re, errbuf, errlen);
257 	m4errx(1, "regular expression error in %s: %s.", source, errbuf);
258 }
259 
260 static void
261 add_sub(int n, const char *string, regex_t *re, regmatch_t *pm)
262 {
263 	if (n > (int)re->re_nsub)
264 		warnx("No subexpression %d", n);
265 	/* Subexpressions that did not match are
266 	 * not an error.  */
267 	else if (pm[n].rm_so != -1 &&
268 	    pm[n].rm_eo != -1) {
269 		addchars(string + pm[n].rm_so,
270 			pm[n].rm_eo - pm[n].rm_so);
271 	}
272 }
273 
274 /* Add replacement string to the output buffer, recognizing special
275  * constructs and replacing them with substrings of the original string.
276  */
277 static void
278 add_replace(const char *string, regex_t *re, const char *replace, regmatch_t *pm)
279 {
280 	const char *p;
281 
282 	for (p = replace; *p != '\0'; p++) {
283 		if (*p == '&' && !mimic_gnu) {
284 			add_sub(0, string, re, pm);
285 			continue;
286 		}
287 		if (*p == '\\') {
288 			if (p[1] == '\\') {
289 				addchar(p[1]);
290 				p++;
291 				continue;
292 			}
293 			if (p[1] == '&') {
294 				if (mimic_gnu)
295 					add_sub(0, string, re, pm);
296 				else
297 					addchar(p[1]);
298 				p++;
299 				continue;
300 			}
301 			if (isdigit((unsigned char)p[1])) {
302 				add_sub(*(++p) - '0', string, re, pm);
303 				continue;
304 			}
305 		}
306 		addchar(*p);
307 	}
308 }
309 
310 static void
311 do_subst(const char *string, regex_t *re, const char *source,
312     const char *replace, regmatch_t *pm)
313 {
314 	int error;
315 	int flags = 0;
316 	const char *last_match = NULL;
317 
318 	while ((error = regexec(re, string, re->re_nsub+1, pm, flags)) == 0) {
319 		if (pm[0].rm_eo != 0) {
320 			if (string[pm[0].rm_eo-1] == '\n')
321 				flags = 0;
322 			else
323 				flags = REG_NOTBOL;
324 		}
325 
326 		/* NULL length matches are special... We use the `vi-mode'
327 		 * rule: don't allow a NULL-match at the last match
328 		 * position.
329 		 */
330 		if (pm[0].rm_so == pm[0].rm_eo &&
331 		    string + pm[0].rm_so == last_match) {
332 			if (*string == '\0')
333 				return;
334 			addchar(*string);
335 			if (*string++ == '\n')
336 				flags = 0;
337 			else
338 				flags = REG_NOTBOL;
339 			continue;
340 		}
341 		last_match = string + pm[0].rm_so;
342 		addchars(string, pm[0].rm_so);
343 		add_replace(string, re, replace, pm);
344 		string += pm[0].rm_eo;
345 	}
346 	if (error != REG_NOMATCH)
347 		exit_regerror(error, re, source);
348 	pbstr(string);
349 }
350 
351 static void
352 do_regexp(const char *string, regex_t *re, const char *source,
353     const char *replace, regmatch_t *pm)
354 {
355 	int error;
356 
357 	switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
358 	case 0:
359 		add_replace(string, re, replace, pm);
360 		pbstr(getstring());
361 		break;
362 	case REG_NOMATCH:
363 		break;
364 	default:
365 		exit_regerror(error, re, source);
366 	}
367 }
368 
369 static void
370 do_regexpindex(const char *string, regex_t *re, const char *source,
371     regmatch_t *pm)
372 {
373 	int error;
374 
375 	switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
376 	case 0:
377 		pbunsigned(pm[0].rm_so);
378 		break;
379 	case REG_NOMATCH:
380 		pbnum(-1);
381 		break;
382 	default:
383 		exit_regerror(error, re, source);
384 	}
385 }
386 
387 /* In Gnu m4 mode, parentheses for backmatch don't work like POSIX 1003.2
388  * says. So we twiddle with the regexp before passing it to regcomp.
389  */
390 static char *
391 twiddle(const char *p)
392 {
393 	/* + at start of regexp is a normal character for Gnu m4 */
394 	if (*p == '^') {
395 		addchar(*p);
396 		p++;
397 	}
398 	if (*p == '+') {
399 		addchar('\\');
400 	}
401 	/* This could use strcspn for speed... */
402 	while (*p != '\0') {
403 		if (*p == '\\') {
404 			switch(p[1]) {
405 			case '(':
406 			case ')':
407 			case '|':
408 				addchar(p[1]);
409 				break;
410 			case 'w':
411 				addconstantstring("[_a-zA-Z0-9]");
412 				break;
413 			case 'W':
414 				addconstantstring("[^_a-zA-Z0-9]");
415 				break;
416 			case '<':
417 				addconstantstring("[[:<:]]");
418 				break;
419 			case '>':
420 				addconstantstring("[[:>:]]");
421 				break;
422 			default:
423 				addchars(p, 2);
424 				break;
425 			}
426 			p+=2;
427 			continue;
428 		}
429 		if (*p == '(' || *p == ')' || *p == '|')
430 			addchar('\\');
431 
432 		addchar(*p);
433 		p++;
434 	}
435 	return getstring();
436 }
437 
438 /* patsubst(string, regexp, opt replacement) */
439 /* argv[2]: string
440  * argv[3]: regexp
441  * argv[4]: opt rep
442  */
443 void
444 dopatsubst(const char *argv[], int argc)
445 {
446 	if (argc <= 3) {
447 		warnx("Too few arguments to patsubst");
448 		return;
449 	}
450 	/* special case: empty regexp */
451 	if (argv[3][0] == '\0') {
452 		const char *s;
453 		size_t len;
454 		if (argc > 4 && argv[4])
455 			len = strlen(argv[4]);
456 		else
457 			len = 0;
458 		for (s = argv[2]; *s != '\0'; s++) {
459 			addchars(argv[4], len);
460 			addchar(*s);
461 		}
462 	} else {
463 		int error;
464 		regex_t re;
465 		regmatch_t *pmatch;
466 		int mode = REG_EXTENDED;
467 		const char *source;
468 		size_t l = strlen(argv[3]);
469 
470 		if (!mimic_gnu ||
471 		    (argv[3][0] == '^') ||
472 		    (l > 0 && argv[3][l-1] == '$'))
473 			mode |= REG_NEWLINE;
474 
475 		source = mimic_gnu ? twiddle(argv[3]) : argv[3];
476 		error = regcomp(&re, source, mode);
477 		if (error != 0)
478 			exit_regerror(error, &re, source);
479 
480 		pmatch = xreallocarray(NULL, re.re_nsub+1, sizeof(regmatch_t),
481 		    NULL);
482 		do_subst(argv[2], &re, source,
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 	const char *source;
497 
498 	if (argc <= 3) {
499 		warnx("Too few arguments to regexp");
500 		return;
501 	}
502 	/* special gnu case */
503 	if (argv[3][0] == '\0' && mimic_gnu) {
504 		if (argc == 4 || argv[4] == NULL)
505 			return;
506 		else
507 			pbstr(argv[4]);
508 	}
509 	source = mimic_gnu ? twiddle(argv[3]) : argv[3];
510 	error = regcomp(&re, source, REG_EXTENDED|REG_NEWLINE);
511 	if (error != 0)
512 		exit_regerror(error, &re, source);
513 
514 	pmatch = xreallocarray(NULL, re.re_nsub+1, sizeof(regmatch_t), NULL);
515 	if (argc == 4 || argv[4] == NULL)
516 		do_regexpindex(argv[2], &re, source, pmatch);
517 	else
518 		do_regexp(argv[2], &re, source, argv[4], pmatch);
519 	free(pmatch);
520 	regfree(&re);
521 }
522 
523 void
524 doformat(const char *argv[], int argc)
525 {
526 	const char *format = argv[2];
527 	int pos = 3;
528 	int left_padded;
529 	long width;
530 	size_t l;
531 	const char *thisarg = NULL;
532 	char temp[2];
533 	long extra;
534 
535 	while (*format != 0) {
536 		if (*format != '%') {
537 			addchar(*format++);
538 			continue;
539 		}
540 
541 		format++;
542 		if (*format == '%') {
543 			addchar(*format++);
544 			continue;
545 		}
546 		if (*format == 0) {
547 			addchar('%');
548 			break;
549 		}
550 
551 		if (*format == '*') {
552 			format++;
553 			if (pos >= argc)
554 				m4errx(1,
555 				    "Format with too many format specifiers.");
556 			width = strtol(argv[pos++], NULL, 10);
557 		} else {
558 			width = strtol(format, __DECONST(char **,&format), 10);
559 		}
560 		if (width < 0) {
561 			left_padded = 1;
562 			width = -width;
563 		} else {
564 			left_padded = 0;
565 		}
566 		if (*format == '.') {
567 			format++;
568 			if (*format == '*') {
569 				format++;
570 				if (pos >= argc)
571 					m4errx(1,
572 					    "Format with too many format specifiers.");
573 				extra = strtol(argv[pos++], NULL, 10);
574 			} else {
575 				extra = strtol(format, __DECONST(char **, &format), 10);
576 			}
577 		} else {
578 			extra = LONG_MAX;
579 		}
580 		if (pos >= argc)
581 			m4errx(1, "Format with too many format specifiers.");
582 		switch(*format) {
583 		case 's':
584 			thisarg = argv[pos++];
585 			break;
586 		case 'c':
587 			temp[0] = strtoul(argv[pos++], NULL, 10);
588 			temp[1] = 0;
589 			thisarg = temp;
590 			break;
591 		default:
592 			m4errx(1, "Unsupported format specification: %s.",
593 			    argv[2]);
594 		}
595 		format++;
596 		l = strlen(thisarg);
597 		if ((long)l > extra)
598 			l = extra;
599 		if (!left_padded) {
600 			while ((long)l < width--)
601 				addchar(' ');
602 		}
603 		addchars(thisarg, l);
604 		if (left_padded) {
605 			while ((long)l < width--)
606 				addchar(' ');
607 		}
608 	}
609 	pbstr(getstring());
610 }
611 
612 void
613 doesyscmd(const char *cmd)
614 {
615 	int p[2];
616 	pid_t pid, cpid;
617 	char *argv[4];
618 	int cc;
619 	int status;
620 
621 	/* Follow gnu m4 documentation: first flush buffers. */
622 	fflush(NULL);
623 
624 	argv[0] = __DECONST(char *, "sh");
625 	argv[1] = __DECONST(char *, "-c");
626 	argv[2] = __DECONST(char *, cmd);
627 	argv[3] = NULL;
628 
629 	/* Just set up standard output, share stderr and stdin with m4 */
630 	if (pipe(p) == -1)
631 		err(1, "bad pipe");
632 	switch(cpid = fork()) {
633 	case -1:
634 		err(1, "bad fork");
635 		/* NOTREACHED */
636 	case 0:
637 		(void) close(p[0]);
638 		(void) dup2(p[1], 1);
639 		(void) close(p[1]);
640 		execv(_PATH_BSHELL, argv);
641 		exit(1);
642 	default:
643 		/* Read result in two stages, since m4's buffer is
644 		 * pushback-only. */
645 		(void) close(p[1]);
646 		do {
647 			char result[BUFSIZE];
648 			cc = read(p[0], result, sizeof result);
649 			if (cc > 0)
650 				addchars(result, cc);
651 		} while (cc > 0 || (cc == -1 && errno == EINTR));
652 
653 		(void) close(p[0]);
654 		while ((pid = wait(&status)) != cpid && pid >= 0)
655 			continue;
656 		pbstr(getstring());
657 	}
658 }
659 
660 void
661 getdivfile(const char *name)
662 {
663 	FILE *f;
664 	int c;
665 
666 	f = fopen(name, "r");
667 	if (!f)
668 		return;
669 
670 	while ((c = getc(f))!= EOF)
671 		putc(c, active);
672 	(void) fclose(f);
673 }
674