1/* $OpenBSD: gnum4.c,v 1.52 2017/08/21 21:41:13 deraadt Exp $ */
2
3/*-
4 * SPDX-License-Identifier: BSD-2-Clause
5 *
6 * Copyright (c) 1999 Marc Espie
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 *    notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 *    notice, this list of conditions and the following disclaimer in the
15 *    documentation and/or other materials provided with the distribution.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 */
29#include <sys/cdefs.h>
30__FBSDID("$FreeBSD: stable/11/usr.bin/m4/gnum4.c 352281 2019-09-13 07:22:09Z bapt $");
31
32/*
33 * functions needed to support gnu-m4 extensions, including a fake freezing
34 */
35
36#include <sys/types.h>
37#include <sys/wait.h>
38#include <ctype.h>
39#include <err.h>
40#include <paths.h>
41#include <regex.h>
42#include <stdarg.h>
43#include <stddef.h>
44#include <stdlib.h>
45#include <stdint.h>
46#include <stdio.h>
47#include <string.h>
48#include <errno.h>
49#include <unistd.h>
50#include <limits.h>
51#include "mdef.h"
52#include "stdd.h"
53#include "extern.h"
54
55
56int mimic_gnu = 0;
57
58/*
59 * Support for include path search
60 * First search in the current directory.
61 * If not found, and the path is not absolute, include path kicks in.
62 * First, -I options, in the order found on the command line.
63 * Then M4PATH env variable
64 */
65
66static struct path_entry {
67	char *name;
68	struct path_entry *next;
69} *first, *last;
70
71static struct path_entry *new_path_entry(const char *);
72static void ensure_m4path(void);
73static struct input_file *dopath(struct input_file *, const char *);
74
75static struct path_entry *
76new_path_entry(const char *dirname)
77{
78	struct path_entry *n;
79
80	n = malloc(sizeof(struct path_entry));
81	if (!n)
82		errx(1, "out of memory");
83	n->name = xstrdup(dirname);
84	n->next = 0;
85	return n;
86}
87
88void
89addtoincludepath(const char *dirname)
90{
91	struct path_entry *n;
92
93	n = new_path_entry(dirname);
94
95	if (last) {
96		last->next = n;
97		last = n;
98	}
99	else
100		last = first = n;
101}
102
103static void
104ensure_m4path(void)
105{
106	static int envpathdone = 0;
107	char *envpath;
108	char *sweep;
109	char *path;
110
111	if (envpathdone)
112		return;
113	envpathdone = TRUE;
114	envpath = getenv("M4PATH");
115	if (!envpath)
116		return;
117	/* for portability: getenv result is read-only */
118	envpath = xstrdup(envpath);
119	for (sweep = envpath;
120	    (path = strsep(&sweep, ":")) != NULL;)
121	    addtoincludepath(path);
122	free(envpath);
123}
124
125static
126struct input_file *
127dopath(struct input_file *i, const char *filename)
128{
129	char path[PATH_MAX];
130	struct path_entry *pe;
131	FILE *f;
132
133	for (pe = first; pe; pe = pe->next) {
134		snprintf(path, sizeof(path), "%s/%s", pe->name, filename);
135		if ((f = fopen(path, "r")) != NULL) {
136			set_input(i, f, path);
137			return i;
138		}
139	}
140	return NULL;
141}
142
143struct input_file *
144fopen_trypath(struct input_file *i, const char *filename)
145{
146	FILE *f;
147
148	f = fopen(filename, "r");
149	if (f != NULL) {
150		set_input(i, f, filename);
151		return i;
152	}
153	if (filename[0] == '/')
154		return NULL;
155
156	ensure_m4path();
157
158	return dopath(i, filename);
159}
160
161void
162doindir(const char *argv[], int argc)
163{
164	ndptr n;
165	struct macro_definition *p = NULL;
166
167	n = lookup(argv[2]);
168	if (n == NULL || (p = macro_getdef(n)) == NULL)
169		m4errx(1, "indir: undefined macro %s.", argv[2]);
170	argv[1] = p->defn;
171
172	eval(argv+1, argc-1, p->type, is_traced(n));
173}
174
175void
176dobuiltin(const char *argv[], int argc)
177{
178	ndptr p;
179
180	argv[1] = NULL;
181	p = macro_getbuiltin(argv[2]);
182	if (p != NULL)
183		eval(argv+1, argc-1, macro_builtin_type(p), is_traced(p));
184	else
185		m4errx(1, "unknown builtin %s.", argv[2]);
186}
187
188
189/* We need some temporary buffer space, as pb pushes BACK and substitution
190 * proceeds forward... */
191static char *buffer;
192static size_t bufsize = 0;
193static size_t current = 0;
194
195static void addchars(const char *, size_t);
196static void addchar(int);
197static char *twiddle(const char *);
198static char *getstring(void);
199static void exit_regerror(int, regex_t *, const char *);
200static void do_subst(const char *, regex_t *, const char *, const char *,
201    regmatch_t *);
202static void do_regexpindex(const char *, regex_t *, const char *, regmatch_t *);
203static void do_regexp(const char *, regex_t *, const char *, const char *,
204    regmatch_t *);
205static void add_sub(int, const char *, regex_t *, regmatch_t *);
206static void add_replace(const char *, regex_t *, const char *, regmatch_t *);
207#define addconstantstring(s) addchars((s), sizeof(s)-1)
208
209static void
210addchars(const char *c, size_t n)
211{
212	if (n == 0)
213		return;
214	while (current + n > bufsize) {
215		if (bufsize == 0)
216			bufsize = 1024;
217		else if (bufsize <= SIZE_MAX/2) {
218			bufsize *= 2;
219		} else {
220			errx(1, "size overflow");
221		}
222		buffer = xrealloc(buffer, bufsize, NULL);
223	}
224	memcpy(buffer+current, c, n);
225	current += n;
226}
227
228static void
229addchar(int c)
230{
231	if (current +1 > bufsize) {
232		if (bufsize == 0)
233			bufsize = 1024;
234		else
235			bufsize *= 2;
236		buffer = xrealloc(buffer, bufsize, NULL);
237	}
238	buffer[current++] = c;
239}
240
241static char *
242getstring(void)
243{
244	addchar('\0');
245	current = 0;
246	return buffer;
247}
248
249
250static void
251exit_regerror(int er, regex_t *re, const char *source)
252{
253	size_t	errlen;
254	char	*errbuf;
255
256	errlen = regerror(er, re, NULL, 0);
257	errbuf = xalloc(errlen,
258	    "malloc in regerror: %lu", (unsigned long)errlen);
259	regerror(er, re, errbuf, errlen);
260	m4errx(1, "regular expression error in %s: %s.", source, errbuf);
261}
262
263/* warnx() plus check to see if we need to change exit code or exit .
264 * -E flag functionality.
265 */
266void
267m4_warnx(const char *fmt, ...)
268{
269	va_list ap;
270
271	va_start(ap, fmt);
272	warnx(fmt, ap);
273	va_end(ap);
274
275	if (fatal_warns)
276		exit(1);
277	if (error_warns)
278		exit_code = 1;
279}
280
281static void
282add_sub(int n, const char *string, regex_t *re, regmatch_t *pm)
283{
284	if (n > (int)re->re_nsub)
285		m4_warnx("No subexpression %d", n);
286	/* Subexpressions that did not match are
287	 * not an error.  */
288	else if (pm[n].rm_so != -1 &&
289	    pm[n].rm_eo != -1) {
290		addchars(string + pm[n].rm_so,
291			pm[n].rm_eo - pm[n].rm_so);
292	}
293}
294
295/* Add replacement string to the output buffer, recognizing special
296 * constructs and replacing them with substrings of the original string.
297 */
298static void
299add_replace(const char *string, regex_t *re, const char *replace, regmatch_t *pm)
300{
301	const char *p;
302
303	for (p = replace; *p != '\0'; p++) {
304		if (*p == '&' && !mimic_gnu) {
305			add_sub(0, string, re, pm);
306			continue;
307		}
308		if (*p == '\\') {
309			if (p[1] == '\\') {
310				addchar(p[1]);
311				p++;
312				continue;
313			}
314			if (p[1] == '&') {
315				if (mimic_gnu)
316					add_sub(0, string, re, pm);
317				else
318					addchar(p[1]);
319				p++;
320				continue;
321			}
322			if (isdigit((unsigned char)p[1])) {
323				add_sub(*(++p) - '0', string, re, pm);
324				continue;
325			}
326		}
327		addchar(*p);
328	}
329}
330
331static void
332do_subst(const char *string, regex_t *re, const char *source,
333    const char *replace, 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, source);
369	pbstr(string);
370}
371
372static void
373do_regexp(const char *string, regex_t *re, const char *source,
374    const char *replace, regmatch_t *pm)
375{
376	int error;
377
378	switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
379	case 0:
380		add_replace(string, re, replace, pm);
381		pbstr(getstring());
382		break;
383	case REG_NOMATCH:
384		break;
385	default:
386		exit_regerror(error, re, source);
387	}
388}
389
390static void
391do_regexpindex(const char *string, regex_t *re, const char *source,
392    regmatch_t *pm)
393{
394	int error;
395
396	switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
397	case 0:
398		pbunsigned(pm[0].rm_so);
399		break;
400	case REG_NOMATCH:
401		pbnum(-1);
402		break;
403	default:
404		exit_regerror(error, re, source);
405	}
406}
407
408/* In Gnu m4 mode, parentheses for backmatch don't work like POSIX 1003.2
409 * says. So we twiddle with the regexp before passing it to regcomp.
410 */
411static char *
412twiddle(const char *p)
413{
414	/* + at start of regexp is a normal character for Gnu m4 */
415	if (*p == '^') {
416		addchar(*p);
417		p++;
418	}
419	if (*p == '+') {
420		addchar('\\');
421	}
422	/* This could use strcspn for speed... */
423	while (*p != '\0') {
424		if (*p == '\\') {
425			switch(p[1]) {
426			case '(':
427			case ')':
428			case '|':
429				addchar(p[1]);
430				break;
431			case 'w':
432				addconstantstring("[_a-zA-Z0-9]");
433				break;
434			case 'W':
435				addconstantstring("[^_a-zA-Z0-9]");
436				break;
437			case '<':
438				addconstantstring("[[:<:]]");
439				break;
440			case '>':
441				addconstantstring("[[:>:]]");
442				break;
443			default:
444				addchars(p, 2);
445				break;
446			}
447			p+=2;
448			continue;
449		}
450		if (*p == '(' || *p == ')' || *p == '|')
451			addchar('\\');
452
453		addchar(*p);
454		p++;
455	}
456	return getstring();
457}
458
459/* patsubst(string, regexp, opt replacement) */
460/* argv[2]: string
461 * argv[3]: regexp
462 * argv[4]: opt rep
463 */
464void
465dopatsubst(const char *argv[], int argc)
466{
467	if (argc <= 3) {
468		m4_warnx("Too few arguments to patsubst");
469		return;
470	}
471	/* special case: empty regexp */
472	if (argv[3][0] == '\0') {
473		const char *s;
474		size_t len;
475		if (argc > 4 && argv[4])
476			len = strlen(argv[4]);
477		else
478			len = 0;
479		for (s = argv[2]; *s != '\0'; s++) {
480			addchars(argv[4], len);
481			addchar(*s);
482		}
483	} else {
484		int error;
485		regex_t re;
486		regmatch_t *pmatch;
487		int mode = REG_EXTENDED;
488		const char *source;
489		size_t l = strlen(argv[3]);
490
491		if (!mimic_gnu ||
492		    (argv[3][0] == '^') ||
493		    (l > 0 && argv[3][l-1] == '$'))
494			mode |= REG_NEWLINE;
495
496		source = mimic_gnu ? twiddle(argv[3]) : argv[3];
497		error = regcomp(&re, source, mode);
498		if (error != 0)
499			exit_regerror(error, &re, source);
500
501		pmatch = xreallocarray(NULL, re.re_nsub+1, sizeof(regmatch_t),
502		    NULL);
503		do_subst(argv[2], &re, source,
504		    argc > 4 && argv[4] != NULL ? argv[4] : "", pmatch);
505		free(pmatch);
506		regfree(&re);
507	}
508	pbstr(getstring());
509}
510
511void
512doregexp(const char *argv[], int argc)
513{
514	int error;
515	regex_t re;
516	regmatch_t *pmatch;
517	const char *source;
518
519	if (argc <= 3) {
520		m4_warnx("Too few arguments to regexp");
521		return;
522	}
523	/* special gnu case */
524	if (argv[3][0] == '\0' && mimic_gnu) {
525		if (argc == 4 || argv[4] == NULL)
526			return;
527		else
528			pbstr(argv[4]);
529	}
530	source = mimic_gnu ? twiddle(argv[3]) : argv[3];
531	error = regcomp(&re, source, REG_EXTENDED|REG_NEWLINE);
532	if (error != 0)
533		exit_regerror(error, &re, source);
534
535	pmatch = xreallocarray(NULL, re.re_nsub+1, sizeof(regmatch_t), NULL);
536	if (argc == 4 || argv[4] == NULL)
537		do_regexpindex(argv[2], &re, source, pmatch);
538	else
539		do_regexp(argv[2], &re, source, argv[4], pmatch);
540	free(pmatch);
541	regfree(&re);
542}
543
544void
545doformat(const char *argv[], int argc)
546{
547	const char *format = argv[2];
548	int pos = 3;
549	int left_padded;
550	long width;
551	size_t l;
552	const char *thisarg = NULL;
553	char temp[2];
554	long extra;
555
556	while (*format != 0) {
557		if (*format != '%') {
558			addchar(*format++);
559			continue;
560		}
561
562		format++;
563		if (*format == '%') {
564			addchar(*format++);
565			continue;
566		}
567		if (*format == 0) {
568			addchar('%');
569			break;
570		}
571
572		if (*format == '*') {
573			format++;
574			if (pos >= argc)
575				m4errx(1,
576				    "Format with too many format specifiers.");
577			width = strtol(argv[pos++], NULL, 10);
578		} else {
579			width = strtol(format, __DECONST(char **,&format), 10);
580		}
581		if (width < 0) {
582			left_padded = 1;
583			width = -width;
584		} else {
585			left_padded = 0;
586		}
587		if (*format == '.') {
588			format++;
589			if (*format == '*') {
590				format++;
591				if (pos >= argc)
592					m4errx(1,
593					    "Format with too many format specifiers.");
594				extra = strtol(argv[pos++], NULL, 10);
595			} else {
596				extra = strtol(format, __DECONST(char **, &format), 10);
597			}
598		} else {
599			extra = LONG_MAX;
600		}
601		if (pos >= argc)
602			m4errx(1, "Format with too many format specifiers.");
603		switch(*format) {
604		case 's':
605			thisarg = argv[pos++];
606			break;
607		case 'c':
608			temp[0] = strtoul(argv[pos++], NULL, 10);
609			temp[1] = 0;
610			thisarg = temp;
611			break;
612		default:
613			m4errx(1, "Unsupported format specification: %s.",
614			    argv[2]);
615		}
616		format++;
617		l = strlen(thisarg);
618		if ((long)l > extra)
619			l = extra;
620		if (!left_padded) {
621			while ((long)l < width--)
622				addchar(' ');
623		}
624		addchars(thisarg, l);
625		if (left_padded) {
626			while ((long)l < width--)
627				addchar(' ');
628		}
629	}
630	pbstr(getstring());
631}
632
633void
634doesyscmd(const char *cmd)
635{
636	int p[2];
637	pid_t cpid;
638	char *argv[4];
639	int cc;
640	int status;
641
642	/* Follow gnu m4 documentation: first flush buffers. */
643	fflush(NULL);
644
645	argv[0] = __DECONST(char *, "sh");
646	argv[1] = __DECONST(char *, "-c");
647	argv[2] = __DECONST(char *, cmd);
648	argv[3] = NULL;
649
650	/* Just set up standard output, share stderr and stdin with m4 */
651	if (pipe(p) == -1)
652		err(1, "bad pipe");
653	switch(cpid = fork()) {
654	case -1:
655		err(1, "bad fork");
656		/* NOTREACHED */
657	case 0:
658		(void) close(p[0]);
659		(void) dup2(p[1], 1);
660		(void) close(p[1]);
661		execv(_PATH_BSHELL, argv);
662		exit(1);
663	default:
664		/* Read result in two stages, since m4's buffer is
665		 * pushback-only. */
666		(void) close(p[1]);
667		do {
668			char result[BUFSIZE];
669			cc = read(p[0], result, sizeof result);
670			if (cc > 0)
671				addchars(result, cc);
672		} while (cc > 0 || (cc == -1 && errno == EINTR));
673
674		(void) close(p[0]);
675		while (waitpid(cpid, &status, 0) == -1) {
676			if (errno != EINTR)
677				break;
678		}
679		pbstr(getstring());
680	}
681}
682
683void
684getdivfile(const char *name)
685{
686	FILE *f;
687	int c;
688
689	f = fopen(name, "r");
690	if (!f)
691		return;
692
693	while ((c = getc(f))!= EOF)
694		putc(c, active);
695	(void) fclose(f);
696}
697