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/*
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 <stdarg.h>
41#include <stddef.h>
42#include <stdlib.h>
43#include <stdint.h>
44#include <stdio.h>
45#include <string.h>
46#include <errno.h>
47#include <unistd.h>
48#include <limits.h>
49#include "mdef.h"
50#include "stdd.h"
51#include "extern.h"
52
53
54int mimic_gnu = 0;
55
56/*
57 * Support for include path search
58 * First search in the current directory.
59 * If not found, and the path is not absolute, include path kicks in.
60 * First, -I options, in the order found on the command line.
61 * Then M4PATH env variable
62 */
63
64static struct path_entry {
65	char *name;
66	struct path_entry *next;
67} *first, *last;
68
69static struct path_entry *new_path_entry(const char *);
70static void ensure_m4path(void);
71static struct input_file *dopath(struct input_file *, const char *);
72
73static struct path_entry *
74new_path_entry(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 = xstrdup(dirname);
82	n->next = 0;
83	return n;
84}
85
86void
87addtoincludepath(const char *dirname)
88{
89	struct path_entry *n;
90
91	n = new_path_entry(dirname);
92
93	if (last) {
94		last->next = n;
95		last = n;
96	}
97	else
98		last = first = n;
99}
100
101static void
102ensure_m4path(void)
103{
104	static int envpathdone = 0;
105	char *envpath;
106	char *sweep;
107	char *path;
108
109	if (envpathdone)
110		return;
111	envpathdone = TRUE;
112	envpath = getenv("M4PATH");
113	if (!envpath)
114		return;
115	/* for portability: getenv result is read-only */
116	envpath = xstrdup(envpath);
117	for (sweep = envpath;
118	    (path = strsep(&sweep, ":")) != NULL;)
119	    addtoincludepath(path);
120	free(envpath);
121}
122
123static
124struct input_file *
125dopath(struct input_file *i, const char *filename)
126{
127	char path[PATH_MAX];
128	struct path_entry *pe;
129	FILE *f;
130
131	for (pe = first; pe; pe = pe->next) {
132		snprintf(path, sizeof(path), "%s/%s", pe->name, filename);
133		if ((f = fopen(path, "r")) != NULL) {
134			set_input(i, f, path);
135			return i;
136		}
137	}
138	return NULL;
139}
140
141struct input_file *
142fopen_trypath(struct input_file *i, const char *filename)
143{
144	FILE *f;
145
146	f = fopen(filename, "r");
147	if (f != NULL) {
148		set_input(i, f, filename);
149		return i;
150	}
151	if (filename[0] == '/')
152		return NULL;
153
154	ensure_m4path();
155
156	return dopath(i, filename);
157}
158
159void
160doindir(const char *argv[], int argc)
161{
162	ndptr n;
163	struct macro_definition *p = NULL;
164
165	n = lookup(argv[2]);
166	if (n == NULL || (p = macro_getdef(n)) == NULL)
167		m4errx(1, "indir: undefined macro %s.", argv[2]);
168	argv[1] = p->defn;
169
170	eval(argv+1, argc-1, p->type, is_traced(n));
171}
172
173void
174dobuiltin(const char *argv[], int argc)
175{
176	ndptr p;
177
178	argv[1] = NULL;
179	p = macro_getbuiltin(argv[2]);
180	if (p != NULL)
181		eval(argv+1, argc-1, macro_builtin_type(p), is_traced(p));
182	else
183		m4errx(1, "unknown builtin %s.", argv[2]);
184}
185
186
187/* We need some temporary buffer space, as pb pushes BACK and substitution
188 * proceeds forward... */
189static char *buffer;
190static size_t bufsize = 0;
191static size_t current = 0;
192
193static void addchars(const char *, size_t);
194static void addchar(int);
195static char *twiddle(const char *);
196static char *getstring(void);
197static void exit_regerror(int, regex_t *, const char *);
198static void do_subst(const char *, regex_t *, const char *, const char *,
199    regmatch_t *);
200static void do_regexpindex(const char *, regex_t *, const char *, regmatch_t *);
201static void do_regexp(const char *, regex_t *, const char *, const char *,
202    regmatch_t *);
203static void add_sub(int, const char *, regex_t *, regmatch_t *);
204static void add_replace(const char *, regex_t *, const char *, regmatch_t *);
205#define addconstantstring(s) addchars((s), sizeof(s)-1)
206
207static void
208addchars(const char *c, size_t n)
209{
210	if (n == 0)
211		return;
212	while (current + n > bufsize) {
213		if (bufsize == 0)
214			bufsize = 1024;
215		else if (bufsize <= SIZE_MAX/2) {
216			bufsize *= 2;
217		} else {
218			errx(1, "size overflow");
219		}
220		buffer = xrealloc(buffer, bufsize, NULL);
221	}
222	memcpy(buffer+current, c, n);
223	current += n;
224}
225
226static void
227addchar(int c)
228{
229	if (current +1 > bufsize) {
230		if (bufsize == 0)
231			bufsize = 1024;
232		else
233			bufsize *= 2;
234		buffer = xrealloc(buffer, bufsize, NULL);
235	}
236	buffer[current++] = c;
237}
238
239static char *
240getstring(void)
241{
242	addchar('\0');
243	current = 0;
244	return buffer;
245}
246
247
248static void
249exit_regerror(int er, regex_t *re, const char *source)
250{
251	size_t	errlen;
252	char	*errbuf;
253
254	errlen = regerror(er, re, NULL, 0);
255	errbuf = xalloc(errlen,
256	    "malloc in regerror: %lu", (unsigned long)errlen);
257	regerror(er, re, errbuf, errlen);
258	m4errx(1, "regular expression error in %s: %s.", source, errbuf);
259}
260
261/* warnx() plus check to see if we need to change exit code or exit .
262 * -E flag functionality.
263 */
264void
265m4_warnx(const char *fmt, ...)
266{
267	va_list ap;
268
269	va_start(ap, fmt);
270	warnx(fmt, ap);
271	va_end(ap);
272
273	if (fatal_warns)
274		exit(1);
275	if (error_warns)
276		exit_code = 1;
277}
278
279static void
280add_sub(int n, const char *string, regex_t *re, regmatch_t *pm)
281{
282	if (n > (int)re->re_nsub)
283		m4_warnx("No subexpression %d", n);
284	/* Subexpressions that did not match are
285	 * not an error.  */
286	else if (pm[n].rm_so != -1 &&
287	    pm[n].rm_eo != -1) {
288		addchars(string + pm[n].rm_so,
289			pm[n].rm_eo - pm[n].rm_so);
290	}
291}
292
293/* Add replacement string to the output buffer, recognizing special
294 * constructs and replacing them with substrings of the original string.
295 */
296static void
297add_replace(const char *string, regex_t *re, const char *replace, regmatch_t *pm)
298{
299	const char *p;
300
301	for (p = replace; *p != '\0'; p++) {
302		if (*p == '&' && !mimic_gnu) {
303			add_sub(0, string, re, pm);
304			continue;
305		}
306		if (*p == '\\') {
307			if (p[1] == '\\') {
308				addchar(p[1]);
309				p++;
310				continue;
311			}
312			if (p[1] == '&') {
313				if (mimic_gnu)
314					add_sub(0, string, re, pm);
315				else
316					addchar(p[1]);
317				p++;
318				continue;
319			}
320			if (isdigit((unsigned char)p[1])) {
321				add_sub(*(++p) - '0', string, re, pm);
322				continue;
323			}
324		}
325		addchar(*p);
326	}
327}
328
329static void
330do_subst(const char *string, regex_t *re, const char *source,
331    const char *replace, regmatch_t *pm)
332{
333	int error;
334	int flags = 0;
335	const char *last_match = NULL;
336
337	while ((error = regexec(re, string, re->re_nsub+1, pm, flags)) == 0) {
338		if (pm[0].rm_eo != 0) {
339			if (string[pm[0].rm_eo-1] == '\n')
340				flags = 0;
341			else
342				flags = REG_NOTBOL;
343		}
344
345		/* NULL length matches are special... We use the `vi-mode'
346		 * rule: don't allow a NULL-match at the last match
347		 * position.
348		 */
349		if (pm[0].rm_so == pm[0].rm_eo &&
350		    string + pm[0].rm_so == last_match) {
351			if (*string == '\0')
352				return;
353			addchar(*string);
354			if (*string++ == '\n')
355				flags = 0;
356			else
357				flags = REG_NOTBOL;
358			continue;
359		}
360		last_match = string + pm[0].rm_so;
361		addchars(string, pm[0].rm_so);
362		add_replace(string, re, replace, pm);
363		string += pm[0].rm_eo;
364	}
365	if (error != REG_NOMATCH)
366		exit_regerror(error, re, source);
367	pbstr(string);
368}
369
370static void
371do_regexp(const char *string, regex_t *re, const char *source,
372    const char *replace, regmatch_t *pm)
373{
374	int error;
375
376	switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
377	case 0:
378		add_replace(string, re, replace, pm);
379		pbstr(getstring());
380		break;
381	case REG_NOMATCH:
382		break;
383	default:
384		exit_regerror(error, re, source);
385	}
386}
387
388static void
389do_regexpindex(const char *string, regex_t *re, const char *source,
390    regmatch_t *pm)
391{
392	int error;
393
394	switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
395	case 0:
396		pbunsigned(pm[0].rm_so);
397		break;
398	case REG_NOMATCH:
399		pbnum(-1);
400		break;
401	default:
402		exit_regerror(error, re, source);
403	}
404}
405
406/* In Gnu m4 mode, parentheses for backmatch don't work like POSIX 1003.2
407 * says. So we twiddle with the regexp before passing it to regcomp.
408 */
409static char *
410twiddle(const char *p)
411{
412	/* + at start of regexp is a normal character for Gnu m4 */
413	if (*p == '^') {
414		addchar(*p);
415		p++;
416	}
417	if (*p == '+') {
418		addchar('\\');
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 */
462void
463dopatsubst(const char *argv[], int argc)
464{
465	if (argc <= 3) {
466		m4_warnx("Too few arguments to patsubst");
467		return;
468	}
469	/* special case: empty regexp */
470	if (argv[3][0] == '\0') {
471		const char *s;
472		size_t len;
473		if (argc > 4 && argv[4])
474			len = strlen(argv[4]);
475		else
476			len = 0;
477		for (s = argv[2]; *s != '\0'; s++) {
478			addchars(argv[4], len);
479			addchar(*s);
480		}
481	} else {
482		int error;
483		regex_t re;
484		regmatch_t *pmatch;
485		int mode = REG_EXTENDED;
486		const char *source;
487		size_t l = strlen(argv[3]);
488
489		if (!mimic_gnu ||
490		    (argv[3][0] == '^') ||
491		    (l > 0 && argv[3][l-1] == '$'))
492			mode |= REG_NEWLINE;
493
494		source = mimic_gnu ? twiddle(argv[3]) : argv[3];
495		error = regcomp(&re, source, mode);
496		if (error != 0)
497			exit_regerror(error, &re, source);
498
499		pmatch = xreallocarray(NULL, re.re_nsub+1, sizeof(regmatch_t),
500		    NULL);
501		do_subst(argv[2], &re, source,
502		    argc > 4 && argv[4] != NULL ? argv[4] : "", pmatch);
503		free(pmatch);
504		regfree(&re);
505	}
506	pbstr(getstring());
507}
508
509void
510doregexp(const char *argv[], int argc)
511{
512	int error;
513	regex_t re;
514	regmatch_t *pmatch;
515	const char *source;
516
517	if (argc <= 3) {
518		m4_warnx("Too few arguments to regexp");
519		return;
520	}
521	/* special gnu case */
522	if (argv[3][0] == '\0' && mimic_gnu) {
523		if (argc == 4 || argv[4] == NULL)
524			return;
525		else
526			pbstr(argv[4]);
527	}
528	source = mimic_gnu ? twiddle(argv[3]) : argv[3];
529	error = regcomp(&re, source, REG_EXTENDED|REG_NEWLINE);
530	if (error != 0)
531		exit_regerror(error, &re, source);
532
533	pmatch = xreallocarray(NULL, re.re_nsub+1, sizeof(regmatch_t), NULL);
534	if (argc == 4 || argv[4] == NULL)
535		do_regexpindex(argv[2], &re, source, pmatch);
536	else
537		do_regexp(argv[2], &re, source, argv[4], pmatch);
538	free(pmatch);
539	regfree(&re);
540}
541
542void
543doformat(const char *argv[], int argc)
544{
545	const char *format = argv[2];
546	int pos = 3;
547	int left_padded;
548	long width;
549	size_t l;
550	const char *thisarg = NULL;
551	char temp[2];
552	long extra;
553
554	while (*format != 0) {
555		if (*format != '%') {
556			addchar(*format++);
557			continue;
558		}
559
560		format++;
561		if (*format == '%') {
562			addchar(*format++);
563			continue;
564		}
565		if (*format == 0) {
566			addchar('%');
567			break;
568		}
569
570		if (*format == '*') {
571			format++;
572			if (pos >= argc)
573				m4errx(1,
574				    "Format with too many format specifiers.");
575			width = strtol(argv[pos++], NULL, 10);
576		} else {
577			width = strtol(format, __DECONST(char **,&format), 10);
578		}
579		if (width < 0) {
580			left_padded = 1;
581			width = -width;
582		} else {
583			left_padded = 0;
584		}
585		if (*format == '.') {
586			format++;
587			if (*format == '*') {
588				format++;
589				if (pos >= argc)
590					m4errx(1,
591					    "Format with too many format specifiers.");
592				extra = strtol(argv[pos++], NULL, 10);
593			} else {
594				extra = strtol(format, __DECONST(char **, &format), 10);
595			}
596		} else {
597			extra = LONG_MAX;
598		}
599		if (pos >= argc)
600			m4errx(1, "Format with too many format specifiers.");
601		switch(*format) {
602		case 's':
603			thisarg = argv[pos++];
604			break;
605		case 'c':
606			temp[0] = strtoul(argv[pos++], NULL, 10);
607			temp[1] = 0;
608			thisarg = temp;
609			break;
610		default:
611			m4errx(1, "Unsupported format specification: %s.",
612			    argv[2]);
613		}
614		format++;
615		l = strlen(thisarg);
616		if ((long)l > extra)
617			l = extra;
618		if (!left_padded) {
619			while ((long)l < width--)
620				addchar(' ');
621		}
622		addchars(thisarg, l);
623		if (left_padded) {
624			while ((long)l < width--)
625				addchar(' ');
626		}
627	}
628	pbstr(getstring());
629}
630
631void
632doesyscmd(const char *cmd)
633{
634	int p[2];
635	pid_t cpid;
636	char *argv[4];
637	int cc;
638	int status;
639
640	/* Follow gnu m4 documentation: first flush buffers. */
641	fflush(NULL);
642
643	argv[0] = __DECONST(char *, "sh");
644	argv[1] = __DECONST(char *, "-c");
645	argv[2] = __DECONST(char *, cmd);
646	argv[3] = NULL;
647
648	/* Just set up standard output, share stderr and stdin with m4 */
649	if (pipe(p) == -1)
650		err(1, "bad pipe");
651	switch(cpid = fork()) {
652	case -1:
653		err(1, "bad fork");
654		/* NOTREACHED */
655	case 0:
656		(void) close(p[0]);
657		(void) dup2(p[1], 1);
658		(void) close(p[1]);
659		execv(_PATH_BSHELL, argv);
660		exit(1);
661	default:
662		/* Read result in two stages, since m4's buffer is
663		 * pushback-only. */
664		(void) close(p[1]);
665		do {
666			char result[BUFSIZE];
667			cc = read(p[0], result, sizeof result);
668			if (cc > 0)
669				addchars(result, cc);
670		} while (cc > 0 || (cc == -1 && errno == EINTR));
671
672		(void) close(p[0]);
673		while (waitpid(cpid, &status, 0) == -1) {
674			if (errno != EINTR)
675				break;
676		}
677		pbstr(getstring());
678	}
679}
680
681void
682getdivfile(const char *name)
683{
684	FILE *f;
685	int c;
686
687	f = fopen(name, "r");
688	if (!f)
689		return;
690
691	while ((c = getc(f))!= EOF)
692		putc(c, active);
693	(void) fclose(f);
694}
695