xref: /netbsd-src/usr.bin/m4/gnum4.c (revision 404fbe5fb94ca1e054339640cabb2801ce52dd30) 
1 /*	$NetBSD: gnum4.c,v 1.5 2004/10/30 20:39:35 dsl Exp $	*/
2 /* $OpenBSD: gnum4.c,v 1.15 2001/10/13 20:18:48 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 #if HAVE_NBTOOL_CONFIG_H
30 #include "nbtool_config.h"
31 #endif
32 
33 /*
34  * functions needed to support gnu-m4 extensions, including a fake freezing
35  */
36 
37 #include <sys/param.h>
38 #include <sys/types.h>
39 #include <sys/wait.h>
40 #include <ctype.h>
41 #include <errno.h>
42 #include <paths.h>
43 #include <regex.h>
44 #include <stddef.h>
45 #include <stdlib.h>
46 #include <stdio.h>
47 #include <string.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 struct path_entry {
64 	char *name;
65 	struct path_entry *next;
66 } *first, *last;
67 
68 static struct path_entry *new_path_entry __P((const char *));
69 static void ensure_m4path __P((void));
70 static struct input_file *dopath __P((struct input_file *, const char *));
71 
72 static struct path_entry *
73 new_path_entry(dirname)
74 	const char *dirname;
75 {
76 	struct path_entry *n;
77 
78 	n = malloc(sizeof(struct path_entry));
79 	if (!n)
80 		errx(1, "out of memory");
81 	n->name = strdup(dirname);
82 	if (!n->name)
83 		errx(1, "out of memory");
84 	n->next = 0;
85 	return n;
86 }
87 
88 void
89 addtoincludepath(dirname)
90 	const char *dirname;
91 {
92 	struct path_entry *n;
93 
94 	n = new_path_entry(dirname);
95 
96 	if (last) {
97 		last->next = n;
98 		last = n;
99 	}
100 	else
101 		last = first = n;
102 }
103 
104 static void
105 ensure_m4path()
106 {
107 	static int envpathdone = 0;
108 	char *envpath;
109 	char *sweep;
110 	char *path;
111 
112 	if (envpathdone)
113 		return;
114 	envpathdone = TRUE;
115 	envpath = getenv("M4PATH");
116 	if (!envpath)
117 		return;
118 	/* for portability: getenv result is read-only */
119 	envpath = strdup(envpath);
120 	if (!envpath)
121 		errx(1, "out of memory");
122 	for (sweep = envpath;
123 	    (path = strsep(&sweep, ":")) != NULL;)
124 	    addtoincludepath(path);
125 	free(envpath);
126 }
127 
128 static
129 struct input_file *
130 dopath(i, filename)
131 	struct input_file *i;
132 	const char *filename;
133 {
134 	char path[MAXPATHLEN];
135 	struct path_entry *pe;
136 	FILE *f;
137 
138 	for (pe = first; pe; pe = pe->next) {
139 		snprintf(path, sizeof(path), "%s/%s", pe->name, filename);
140 		if ((f = fopen(path, "r")) != 0) {
141 			set_input(i, f, path);
142 			return i;
143 		}
144 	}
145 	return NULL;
146 }
147 
148 struct input_file *
149 fopen_trypath(i, filename)
150 	struct input_file *i;
151 	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(argv, argc)
170 	const char *argv[];
171 	int argc;
172 {
173 	ndptr p;
174 
175 	p = lookup(argv[2]);
176 	if (p == NULL)
177 		errx(1, "undefined macro %s", argv[2]);
178 	argv[1] = p->defn;
179 	eval(argv+1, argc-1, p->type);
180 }
181 
182 void
183 dobuiltin(argv, argc)
184 	const char *argv[];
185 	int argc;
186 {
187 	int n;
188 	argv[1] = NULL;
189 	n = builtin_type(argv[2]);
190 	if (n != -1)
191 		eval(argv+1, argc-1, n);
192 	else
193 		errx(1, "unknown builtin %s", argv[2]);
194 }
195 
196 
197 /* We need some temporary buffer space, as pb pushes BACK and substitution
198  * proceeds forward... */
199 static char *buffer;
200 static size_t bufsize = 0;
201 static size_t current = 0;
202 
203 static void addchars __P((const char *, size_t));
204 static void addchar __P((char));
205 static char *twiddle __P((const char *));
206 static char *getstring __P((void));
207 static void exit_regerror __P((int, regex_t *));
208 static void do_subst __P((const char *, regex_t *, const char *, regmatch_t *));
209 static void do_regexpindex __P((const char *, regex_t *, regmatch_t *));
210 static void do_regexp __P((const char *, regex_t *, const char *, regmatch_t *));
211 static void add_sub __P((int, const char *, regex_t *, regmatch_t *));
212 static void add_replace __P((const char *, regex_t *, const char *, regmatch_t *));
213 #define addconstantstring(s) addchars((s), sizeof(s)-1)
214 
215 static void
216 addchars(c, n)
217 	const char *c;
218 	size_t n;
219 {
220 	if (n == 0)
221 		return;
222 	while (current + n > bufsize) {
223 		if (bufsize == 0)
224 			bufsize = 1024;
225 		else
226 			bufsize *= 2;
227 		buffer = realloc(buffer, bufsize);
228 		if (buffer == NULL)
229 			errx(1, "out of memory");
230 	}
231 	memcpy(buffer+current, c, n);
232 	current += n;
233 }
234 
235 static void
236 addchar(c)
237 	char c;
238 {
239 	if (current +1 > bufsize) {
240 		if (bufsize == 0)
241 			bufsize = 1024;
242 		else
243 			bufsize *= 2;
244 		buffer = realloc(buffer, bufsize);
245 		if (buffer == NULL)
246 			errx(1, "out of memory");
247 	}
248 	buffer[current++] = c;
249 }
250 
251 static char *
252 getstring()
253 {
254 	addchar('\0');
255 	current = 0;
256 	return buffer;
257 }
258 
259 
260 static void
261 exit_regerror(er, re)
262 	int er;
263 	regex_t *re;
264 {
265 	size_t 	errlen;
266 	char 	*errbuf;
267 
268 	errlen = regerror(er, re, NULL, 0);
269 	errbuf = xalloc(errlen);
270 	regerror(er, re, errbuf, errlen);
271 	errx(1, "regular expression error: %s", errbuf);
272 }
273 
274 static void
275 add_sub(n, string, re, pm)
276 	int n;
277 	const char *string;
278 	regex_t *re;
279 	regmatch_t *pm;
280 {
281 	if (n > re->re_nsub)
282 		warnx("No subexpression %d", n);
283 	/* Subexpressions that did not match are
284 	 * not an error.  */
285 	else if (pm[n].rm_so != -1 &&
286 	    pm[n].rm_eo != -1) {
287 		addchars(string + pm[n].rm_so,
288 			pm[n].rm_eo - pm[n].rm_so);
289 	}
290 }
291 
292 /* Add replacement string to the output buffer, recognizing special
293  * constructs and replacing them with substrings of the original string.
294  */
295 static void
296 add_replace(string, re, replace, pm)
297 	const char *string;
298 	regex_t *re;
299 	const char *replace;
300 	regmatch_t *pm;
301 {
302 	const char *p;
303 
304 	for (p = replace; *p != '\0'; p++) {
305 		if (*p == '&' && !mimic_gnu) {
306 			add_sub(0, string, re, pm);
307 			continue;
308 		}
309 		if (*p == '\\') {
310 			if (p[1] == '\\') {
311 				addchar(p[1]);
312 				p++;
313 				continue;
314 			}
315 			if (p[1] == '&') {
316 				if (mimic_gnu)
317 					add_sub(0, string, re, pm);
318 				else
319 					addchar(p[1]);
320 				p++;
321 				continue;
322 			}
323 			if (isdigit((unsigned char)p[1])) {
324 				add_sub(*(++p) - '0', string, re, pm);
325 				continue;
326 			}
327 		}
328 	    	addchar(*p);
329 	}
330 }
331 
332 static void
333 do_subst(string, re, replace, pm)
334 	const char *string;
335 	regex_t *re;
336 	const char *replace;
337 	regmatch_t *pm;
338 {
339 	int error;
340 	int flags = 0;
341 	const char *last_match = NULL;
342 
343 	while ((error = regexec(re, string, re->re_nsub+1, pm, flags)) == 0) {
344 		if (pm[0].rm_eo != 0) {
345 			if (string[pm[0].rm_eo-1] == '\n')
346 				flags = 0;
347 			else
348 				flags = REG_NOTBOL;
349 		}
350 
351 		/* NULL length matches are special... We use the `vi-mode'
352 		 * rule: don't allow a NULL-match at the last match
353 		 * position.
354 		 */
355 		if (pm[0].rm_so == pm[0].rm_eo &&
356 		    string + pm[0].rm_so == last_match) {
357 			if (*string == '\0')
358 				return;
359 			addchar(*string);
360 			if (*string++ == '\n')
361 				flags = 0;
362 			else
363 				flags = REG_NOTBOL;
364 			continue;
365 		}
366 		last_match = string + pm[0].rm_so;
367 		addchars(string, pm[0].rm_so);
368 		add_replace(string, re, replace, pm);
369 		string += pm[0].rm_eo;
370 	}
371 	if (error != REG_NOMATCH)
372 		exit_regerror(error, re);
373 	pbstr(string);
374 }
375 
376 static void
377 do_regexp(string, re, replace, pm)
378 	const char *string;
379 	regex_t *re;
380 	const char *replace;
381 	regmatch_t *pm;
382 {
383 	int error;
384 
385 	switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
386 	case 0:
387 		add_replace(string, re, replace, pm);
388 		pbstr(getstring());
389 		break;
390 	case REG_NOMATCH:
391 		break;
392 	default:
393 		exit_regerror(error, re);
394 	}
395 }
396 
397 static void
398 do_regexpindex(string, re, pm)
399 	const char *string;
400 	regex_t *re;
401 	regmatch_t *pm;
402 {
403 	int error;
404 
405 	switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
406 	case 0:
407 		pbunsigned(pm[0].rm_so);
408 		break;
409 	case REG_NOMATCH:
410 		pbnum(-1);
411 		break;
412 	default:
413 		exit_regerror(error, re);
414 	}
415 }
416 
417 /* In Gnu m4 mode, parentheses for backmatch don't work like POSIX 1003.2
418  * says. So we twiddle with the regexp before passing it to regcomp.
419  */
420 static char *
421 twiddle(p)
422 	const char *p;
423 {
424 	/* This could use strcspn for speed... */
425 	while (*p != '\0') {
426 		if (*p == '\\') {
427 			switch(p[1]) {
428 			case '(':
429 			case ')':
430 			case '|':
431 				addchar(p[1]);
432 				break;
433 			case 'w':
434 				addconstantstring("[_a-zA-Z0-9]");
435 				break;
436 			case 'W':
437 				addconstantstring("[^_a-zA-Z0-9]");
438 				break;
439 			case '<':
440 				addconstantstring("[[:<:]]");
441 				break;
442 			case '>':
443 				addconstantstring("[[:>:]]");
444 				break;
445 			default:
446 				addchars(p, 2);
447 				break;
448 			}
449 			p+=2;
450 			continue;
451 		}
452 		if (*p == '(' || *p == ')' || *p == '|')
453 			addchar('\\');
454 
455 		addchar(*p);
456 		p++;
457 	}
458 	return getstring();
459 }
460 
461 /* patsubst(string, regexp, opt replacement) */
462 /* argv[2]: string
463  * argv[3]: regexp
464  * argv[4]: opt rep
465  */
466 void
467 dopatsubst(argv, argc)
468 	const char *argv[];
469 	int argc;
470 {
471 	int error;
472 	regex_t re;
473 	regmatch_t *pmatch;
474 
475 	if (argc <= 3) {
476 		warnx("Too few arguments to patsubst");
477 		return;
478 	}
479 	error = regcomp(&re, mimic_gnu ? twiddle(argv[3]) : argv[3],
480 	    REG_NEWLINE | REG_EXTENDED);
481 	if (error != 0)
482 		exit_regerror(error, &re);
483 
484 	pmatch = xalloc(sizeof(regmatch_t) * (re.re_nsub+1));
485 	do_subst(argv[2], &re,
486 	    argc != 4 && argv[4] != NULL ? argv[4] : "", pmatch);
487 	pbstr(getstring());
488 	free(pmatch);
489 	regfree(&re);
490 }
491 
492 void
493 doregexp(argv, argc)
494 	const char *argv[];
495 	int argc;
496 {
497 	int error;
498 	regex_t re;
499 	regmatch_t *pmatch;
500 
501 	if (argc <= 3) {
502 		warnx("Too few arguments to regexp");
503 		return;
504 	}
505 	error = regcomp(&re, mimic_gnu ? twiddle(argv[3]) : argv[3],
506 	    REG_EXTENDED);
507 	if (error != 0)
508 		exit_regerror(error, &re);
509 
510 	pmatch = xalloc(sizeof(regmatch_t) * (re.re_nsub+1));
511 	if (argv[4] == NULL || argc == 4)
512 		do_regexpindex(argv[2], &re, pmatch);
513 	else
514 		do_regexp(argv[2], &re, argv[4], pmatch);
515 	free(pmatch);
516 	regfree(&re);
517 }
518 
519 void
520 doesyscmd(cmd)
521 	const char *cmd;
522 {
523 	int p[2];
524 	pid_t pid, cpid;
525 	char *argv[4];
526 	int cc;
527 	int status;
528 
529 	/* Follow gnu m4 documentation: first flush buffers. */
530 	fflush(NULL);
531 
532 	argv[0] = "sh";
533 	argv[1] = "-c";
534 	argv[2] = (char *)cmd;
535 	argv[3] = NULL;
536 
537 	/* Just set up standard output, share stderr and stdin with m4 */
538 	if (pipe(p) == -1)
539 		err(1, "bad pipe");
540 	switch(cpid = fork()) {
541 	case -1:
542 		err(1, "bad fork");
543 		/* NOTREACHED */
544 	case 0:
545 		(void) close(p[0]);
546 		(void) dup2(p[1], 1);
547 		(void) close(p[1]);
548 		execv(_PATH_BSHELL, argv);
549 		exit(1);
550 	default:
551 		/* Read result in two stages, since m4's buffer is
552 		 * pushback-only. */
553 		(void) close(p[1]);
554 		do {
555 			char result[BUFSIZE];
556 			cc = read(p[0], result, sizeof result);
557 			if (cc > 0)
558 				addchars(result, cc);
559 		} while (cc > 0 || (cc == -1 && errno == EINTR));
560 
561 		(void) close(p[0]);
562 		while ((pid = wait(&status)) != cpid && pid >= 0)
563 			continue;
564 		pbstr(getstring());
565 	}
566 }
567