xref: /netbsd/usr.bin/m4/gnum4.c (revision bf9ec67e) 
1 /*	$NetBSD: gnum4.c,v 1.2 2002/01/21 21:49:57 tv 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 /*
30  * functions needed to support gnu-m4 extensions, including a fake freezing
31  */
32 
33 #include <sys/param.h>
34 #include <sys/types.h>
35 #include <sys/wait.h>
36 #include <ctype.h>
37 #include <errno.h>
38 #include <paths.h>
39 #include <regex.h>
40 #include <stddef.h>
41 #include <stdlib.h>
42 #include <stdio.h>
43 #include <string.h>
44 #include "mdef.h"
45 #include "stdd.h"
46 #include "extern.h"
47 
48 
49 int mimic_gnu = 0;
50 
51 /*
52  * Support for include path search
53  * First search in the the current directory.
54  * If not found, and the path is not absolute, include path kicks in.
55  * First, -I options, in the order found on the command line.
56  * Then M4PATH env variable
57  */
58 
59 struct path_entry {
60 	char *name;
61 	struct path_entry *next;
62 } *first, *last;
63 
64 static struct path_entry *new_path_entry __P((const char *));
65 static void ensure_m4path __P((void));
66 static struct input_file *dopath __P((struct input_file *, const char *));
67 
68 static struct path_entry *
69 new_path_entry(dirname)
70 	const char *dirname;
71 {
72 	struct path_entry *n;
73 
74 	n = malloc(sizeof(struct path_entry));
75 	if (!n)
76 		errx(1, "out of memory");
77 	n->name = strdup(dirname);
78 	if (!n->name)
79 		errx(1, "out of memory");
80 	n->next = 0;
81 	return n;
82 }
83 
84 void
85 addtoincludepath(dirname)
86 	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()
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 = strdup(envpath);
116 	if (!envpath)
117 		errx(1, "out of memory");
118 	for (sweep = envpath;
119 	    (path = strsep(&sweep, ":")) != NULL;)
120 	    addtoincludepath(path);
121 	free(envpath);
122 }
123 
124 static
125 struct input_file *
126 dopath(i, filename)
127 	struct input_file *i;
128 	const char *filename;
129 {
130 	char path[MAXPATHLEN];
131 	struct path_entry *pe;
132 	FILE *f;
133 
134 	for (pe = first; pe; pe = pe->next) {
135 		snprintf(path, sizeof(path), "%s/%s", pe->name, filename);
136 		if ((f = fopen(path, "r")) != 0) {
137 			set_input(i, f, path);
138 			return i;
139 		}
140 	}
141 	return NULL;
142 }
143 
144 struct input_file *
145 fopen_trypath(i, filename)
146 	struct input_file *i;
147 	const char *filename;
148 {
149 	FILE *f;
150 
151 	f = fopen(filename, "r");
152 	if (f != NULL) {
153 		set_input(i, f, filename);
154 		return i;
155 	}
156 	if (filename[0] == '/')
157 		return NULL;
158 
159 	ensure_m4path();
160 
161 	return dopath(i, filename);
162 }
163 
164 void
165 doindir(argv, argc)
166 	const char *argv[];
167 	int argc;
168 {
169 	ndptr p;
170 
171 	p = lookup(argv[2]);
172 	if (p == NULL)
173 		errx(1, "undefined macro %s", argv[2]);
174 	argv[1] = p->defn;
175 	eval(argv+1, argc-1, p->type);
176 }
177 
178 void
179 dobuiltin(argv, argc)
180 	const char *argv[];
181 	int argc;
182 {
183 	int n;
184 	argv[1] = NULL;
185 	n = builtin_type(argv[2]);
186 	if (n != -1)
187 		eval(argv+1, argc-1, n);
188 	else
189 		errx(1, "unknown builtin %s", argv[2]);
190 }
191 
192 
193 /* We need some temporary buffer space, as pb pushes BACK and substitution
194  * proceeds forward... */
195 static char *buffer;
196 static size_t bufsize = 0;
197 static size_t current = 0;
198 
199 static void addchars __P((const char *, size_t));
200 static void addchar __P((char));
201 static char *twiddle __P((const char *));
202 static char *getstring __P((void));
203 static void exit_regerror __P((int, regex_t *));
204 static void do_subst __P((const char *, regex_t *, const char *, regmatch_t *));
205 static void do_regexpindex __P((const char *, regex_t *, regmatch_t *));
206 static void do_regexp __P((const char *, regex_t *, const char *, regmatch_t *));
207 static void add_sub __P((int, const char *, regex_t *, regmatch_t *));
208 static void add_replace __P((const char *, regex_t *, const char *, regmatch_t *));
209 #define addconstantstring(s) addchars((s), sizeof(s)-1)
210 
211 static void
212 addchars(c, n)
213 	const char *c;
214 	size_t n;
215 {
216 	if (n == 0)
217 		return;
218 	while (current + n > bufsize) {
219 		if (bufsize == 0)
220 			bufsize = 1024;
221 		else
222 			bufsize *= 2;
223 		buffer = realloc(buffer, bufsize);
224 		if (buffer == NULL)
225 			errx(1, "out of memory");
226 	}
227 	memcpy(buffer+current, c, n);
228 	current += n;
229 }
230 
231 static void
232 addchar(c)
233 	char c;
234 {
235 	if (current +1 > bufsize) {
236 		if (bufsize == 0)
237 			bufsize = 1024;
238 		else
239 			bufsize *= 2;
240 		buffer = realloc(buffer, bufsize);
241 		if (buffer == NULL)
242 			errx(1, "out of memory");
243 	}
244 	buffer[current++] = c;
245 }
246 
247 static char *
248 getstring()
249 {
250 	addchar('\0');
251 	current = 0;
252 	return buffer;
253 }
254 
255 
256 static void
257 exit_regerror(er, re)
258 	int er;
259 	regex_t *re;
260 {
261 	size_t 	errlen;
262 	char 	*errbuf;
263 
264 	errlen = regerror(er, re, NULL, 0);
265 	errbuf = xalloc(errlen);
266 	regerror(er, re, errbuf, errlen);
267 	errx(1, "regular expression error: %s", errbuf);
268 }
269 
270 static void
271 add_sub(n, string, re, pm)
272 	int n;
273 	const char *string;
274 	regex_t *re;
275 	regmatch_t *pm;
276 {
277 	if (n > re->re_nsub)
278 		warnx("No subexpression %d", n);
279 	/* Subexpressions that did not match are
280 	 * not an error.  */
281 	else if (pm[n].rm_so != -1 &&
282 	    pm[n].rm_eo != -1) {
283 		addchars(string + pm[n].rm_so,
284 			pm[n].rm_eo - pm[n].rm_so);
285 	}
286 }
287 
288 /* Add replacement string to the output buffer, recognizing special
289  * constructs and replacing them with substrings of the original string.
290  */
291 static void
292 add_replace(string, re, replace, pm)
293 	const char *string;
294 	regex_t *re;
295 	const char *replace;
296 	regmatch_t *pm;
297 {
298 	const char *p;
299 
300 	for (p = replace; *p != '\0'; p++) {
301 		if (*p == '&' && !mimic_gnu) {
302 			add_sub(0, string, re, pm);
303 			continue;
304 		}
305 		if (*p == '\\') {
306 			if (p[1] == '\\') {
307 				addchar(p[1]);
308 				p++;
309 				continue;
310 			}
311 			if (p[1] == '&') {
312 				if (mimic_gnu)
313 					add_sub(0, string, re, pm);
314 				else
315 					addchar(p[1]);
316 				p++;
317 				continue;
318 			}
319 			if (isdigit(p[1])) {
320 				add_sub(*(++p) - '0', string, re, pm);
321 				continue;
322 			}
323 		}
324 	    	addchar(*p);
325 	}
326 }
327 
328 static void
329 do_subst(string, re, replace, pm)
330 	const char *string;
331 	regex_t *re;
332 	const char *replace;
333 	regmatch_t *pm;
334 {
335 	int error;
336 	int flags = 0;
337 	const char *last_match = NULL;
338 
339 	while ((error = regexec(re, string, re->re_nsub+1, pm, flags)) == 0) {
340 		if (pm[0].rm_eo != 0) {
341 			if (string[pm[0].rm_eo-1] == '\n')
342 				flags = 0;
343 			else
344 				flags = REG_NOTBOL;
345 		}
346 
347 		/* NULL length matches are special... We use the `vi-mode'
348 		 * rule: don't allow a NULL-match at the last match
349 		 * position.
350 		 */
351 		if (pm[0].rm_so == pm[0].rm_eo &&
352 		    string + pm[0].rm_so == last_match) {
353 			if (*string == '\0')
354 				return;
355 			addchar(*string);
356 			if (*string++ == '\n')
357 				flags = 0;
358 			else
359 				flags = REG_NOTBOL;
360 			continue;
361 		}
362 		last_match = string + pm[0].rm_so;
363 		addchars(string, pm[0].rm_so);
364 		add_replace(string, re, replace, pm);
365 		string += pm[0].rm_eo;
366 	}
367 	if (error != REG_NOMATCH)
368 		exit_regerror(error, re);
369 	pbstr(string);
370 }
371 
372 static void
373 do_regexp(string, re, replace, pm)
374 	const char *string;
375 	regex_t *re;
376 	const char *replace;
377 	regmatch_t *pm;
378 {
379 	int error;
380 
381 	switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
382 	case 0:
383 		add_replace(string, re, replace, pm);
384 		pbstr(getstring());
385 		break;
386 	case REG_NOMATCH:
387 		break;
388 	default:
389 		exit_regerror(error, re);
390 	}
391 }
392 
393 static void
394 do_regexpindex(string, re, pm)
395 	const char *string;
396 	regex_t *re;
397 	regmatch_t *pm;
398 {
399 	int error;
400 
401 	switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
402 	case 0:
403 		pbunsigned(pm[0].rm_so);
404 		break;
405 	case REG_NOMATCH:
406 		pbnum(-1);
407 		break;
408 	default:
409 		exit_regerror(error, re);
410 	}
411 }
412 
413 /* In Gnu m4 mode, parentheses for backmatch don't work like POSIX 1003.2
414  * says. So we twiddle with the regexp before passing it to regcomp.
415  */
416 static char *
417 twiddle(p)
418 	const char *p;
419 {
420 	/* This could use strcspn for speed... */
421 	while (*p != '\0') {
422 		if (*p == '\\') {
423 			switch(p[1]) {
424 			case '(':
425 			case ')':
426 			case '|':
427 				addchar(p[1]);
428 				break;
429 			case 'w':
430 				addconstantstring("[_a-zA-Z0-9]");
431 				break;
432 			case 'W':
433 				addconstantstring("[^_a-zA-Z0-9]");
434 				break;
435 			case '<':
436 				addconstantstring("[[:<:]]");
437 				break;
438 			case '>':
439 				addconstantstring("[[:>:]]");
440 				break;
441 			default:
442 				addchars(p, 2);
443 				break;
444 			}
445 			p+=2;
446 			continue;
447 		}
448 		if (*p == '(' || *p == ')' || *p == '|')
449 			addchar('\\');
450 
451 		addchar(*p);
452 		p++;
453 	}
454 	return getstring();
455 }
456 
457 /* patsubst(string, regexp, opt replacement) */
458 /* argv[2]: string
459  * argv[3]: regexp
460  * argv[4]: opt rep
461  */
462 void
463 dopatsubst(argv, argc)
464 	const char *argv[];
465 	int argc;
466 {
467 	int error;
468 	regex_t re;
469 	regmatch_t *pmatch;
470 
471 	if (argc <= 3) {
472 		warnx("Too few arguments to patsubst");
473 		return;
474 	}
475 	error = regcomp(&re, mimic_gnu ? twiddle(argv[3]) : argv[3],
476 	    REG_NEWLINE | REG_EXTENDED);
477 	if (error != 0)
478 		exit_regerror(error, &re);
479 
480 	pmatch = xalloc(sizeof(regmatch_t) * (re.re_nsub+1));
481 	do_subst(argv[2], &re,
482 	    argc != 4 && argv[4] != NULL ? argv[4] : "", pmatch);
483 	pbstr(getstring());
484 	free(pmatch);
485 	regfree(&re);
486 }
487 
488 void
489 doregexp(argv, argc)
490 	const char *argv[];
491 	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));
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 doesyscmd(cmd)
517 	const char *cmd;
518 {
519 	int p[2];
520 	pid_t pid, cpid;
521 	char *argv[4];
522 	int cc;
523 	int status;
524 
525 	/* Follow gnu m4 documentation: first flush buffers. */
526 	fflush(NULL);
527 
528 	argv[0] = "sh";
529 	argv[1] = "-c";
530 	argv[2] = (char *)cmd;
531 	argv[3] = NULL;
532 
533 	/* Just set up standard output, share stderr and stdin with m4 */
534 	if (pipe(p) == -1)
535 		err(1, "bad pipe");
536 	switch(cpid = fork()) {
537 	case -1:
538 		err(1, "bad fork");
539 		/* NOTREACHED */
540 	case 0:
541 		(void) close(p[0]);
542 		(void) dup2(p[1], 1);
543 		(void) close(p[1]);
544 		execv(_PATH_BSHELL, argv);
545 		exit(1);
546 	default:
547 		/* Read result in two stages, since m4's buffer is
548 		 * pushback-only. */
549 		(void) close(p[1]);
550 		do {
551 			char result[BUFSIZE];
552 			cc = read(p[0], result, sizeof result);
553 			if (cc > 0)
554 				addchars(result, cc);
555 		} while (cc > 0 || (cc == -1 && errno == EINTR));
556 
557 		(void) close(p[0]);
558 		while ((pid = wait(&status)) != cpid && pid >= 0)
559 			continue;
560 		pbstr(getstring());
561 	}
562 }
563