xref: /freebsd/usr.bin/m4/main.c (revision e17f5b1d)
1 /*	$OpenBSD: main.c,v 1.87 2017/06/15 13:48:42 bcallah Exp $	*/
2 /*	$NetBSD: main.c,v 1.12 1997/02/08 23:54:49 cgd Exp $	*/
3 
4 /*-
5  * SPDX-License-Identifier: BSD-3-Clause
6  *
7  * Copyright (c) 1989, 1993
8  *	The Regents of the University of California.  All rights reserved.
9  *
10  * This code is derived from software contributed to Berkeley by
11  * Ozan Yigit at York University.
12  *
13  * Redistribution and use in source and binary forms, with or without
14  * modification, are permitted provided that the following conditions
15  * are met:
16  * 1. Redistributions of source code must retain the above copyright
17  *    notice, this list of conditions and the following disclaimer.
18  * 2. Redistributions in binary form must reproduce the above copyright
19  *    notice, this list of conditions and the following disclaimer in the
20  *    documentation and/or other materials provided with the distribution.
21  * 3. Neither the name of the University nor the names of its contributors
22  *    may be used to endorse or promote products derived from this software
23  *    without specific prior written permission.
24  *
25  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
34  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35  * SUCH DAMAGE.
36  */
37 
38 /*
39  * main.c
40  * Facility: m4 macro processor
41  * by: oz
42  */
43 #include <sys/cdefs.h>
44 __FBSDID("$FreeBSD$");
45 
46 #include <assert.h>
47 #include <signal.h>
48 #include <err.h>
49 #include <errno.h>
50 #include <unistd.h>
51 #include <stdio.h>
52 #include <ctype.h>
53 #include <string.h>
54 #include <stddef.h>
55 #include <stdint.h>
56 #include <stdlib.h>
57 #include <ohash.h>
58 #include "mdef.h"
59 #include "stdd.h"
60 #include "extern.h"
61 #include "pathnames.h"
62 
63 stae *mstack;			/* stack of m4 machine         */
64 char *sstack;			/* shadow stack, for string space extension */
65 static size_t STACKMAX;		/* current maximum size of stack */
66 int sp;				/* current m4  stack pointer   */
67 int fp;				/* m4 call frame pointer       */
68 struct input_file infile[MAXINP];/* input file stack (0=stdin)  */
69 FILE **outfile;			/* diversion array(0=bitbucket)*/
70 int maxout;
71 FILE *active;			/* active output file pointer  */
72 int ilevel = 0;			/* input file stack pointer    */
73 int oindex = 0;			/* diversion index..	       */
74 const char *null = "";                /* as it says.. just a null..  */
75 char **m4wraps = NULL;		/* m4wraps array.	       */
76 int maxwraps = 0;		/* size of m4wraps array       */
77 int wrapindex = 0;		/* current offset in m4wraps   */
78 char lquote[MAXCCHARS+1] = {LQUOTE};	/* left quote character  (`)   */
79 char rquote[MAXCCHARS+1] = {RQUOTE};	/* right quote character (')   */
80 char scommt[MAXCCHARS+1] = {SCOMMT};	/* start character for comment */
81 char ecommt[MAXCCHARS+1] = {ECOMMT};	/* end character for comment   */
82 int  synch_lines = 0;		/* line synchronisation for C preprocessor */
83 int  prefix_builtins = 0;	/* -P option to prefix builtin keywords */
84 int  error_warns = 0;		/* -E option to make warnings exit_code = 1 */
85 int  fatal_warns = 0;		/* -E -E option to make warnings fatal */
86 
87 struct keyblk {
88         const char    *knam;          /* keyword name */
89         int     ktyp;           /* keyword type */
90 };
91 
92 static struct keyblk keywrds[] = {	/* m4 keywords to be installed */
93 	{ "include",      INCLTYPE },
94 	{ "sinclude",     SINCTYPE },
95 	{ "define",       DEFITYPE },
96 	{ "defn",         DEFNTYPE },
97 	{ "divert",       DIVRTYPE | NOARGS },
98 	{ "expr",         EXPRTYPE },
99 	{ "eval",         EXPRTYPE },
100 	{ "substr",       SUBSTYPE },
101 	{ "ifelse",       IFELTYPE },
102 	{ "ifdef",        IFDFTYPE },
103 	{ "len",          LENGTYPE },
104 	{ "incr",         INCRTYPE },
105 	{ "decr",         DECRTYPE },
106 	{ "dnl",          DNLNTYPE | NOARGS },
107 	{ "changequote",  CHNQTYPE | NOARGS },
108 	{ "changecom",    CHNCTYPE | NOARGS },
109 	{ "index",        INDXTYPE },
110 #ifdef EXTENDED
111 	{ "paste",        PASTTYPE },
112 	{ "spaste",       SPASTYPE },
113 	/* Newer extensions, needed to handle gnu-m4 scripts */
114 	{ "indir",        INDIRTYPE},
115 	{ "builtin",      BUILTINTYPE},
116 	{ "patsubst",	  PATSTYPE},
117 	{ "regexp",	  REGEXPTYPE},
118 	{ "esyscmd",	  ESYSCMDTYPE},
119 	{ "__file__",	  FILENAMETYPE | NOARGS},
120 	{ "__line__",	  LINETYPE | NOARGS},
121 #endif
122 	{ "popdef",       POPDTYPE },
123 	{ "pushdef",      PUSDTYPE },
124 	{ "dumpdef",      DUMPTYPE | NOARGS },
125 	{ "shift",        SHIFTYPE | NOARGS },
126 	{ "translit",     TRNLTYPE },
127 	{ "undefine",     UNDFTYPE },
128 	{ "undivert",     UNDVTYPE | NOARGS },
129 	{ "divnum",       DIVNTYPE | NOARGS },
130 	{ "maketemp",     MKTMTYPE },
131 	{ "mkstemp",      MKTMTYPE },
132 	{ "errprint",     ERRPTYPE | NOARGS },
133 	{ "m4wrap",       M4WRTYPE | NOARGS },
134 	{ "m4exit",       EXITTYPE | NOARGS },
135 	{ "syscmd",       SYSCTYPE },
136 	{ "sysval",       SYSVTYPE | NOARGS },
137 	{ "traceon",	  TRACEONTYPE | NOARGS },
138 	{ "traceoff",	  TRACEOFFTYPE | NOARGS },
139 
140 	{ "unix",         SELFTYPE | NOARGS },
141 };
142 
143 #define MAXKEYS	(sizeof(keywrds)/sizeof(struct keyblk))
144 
145 extern int optind;
146 extern char *optarg;
147 
148 #define MAXRECORD 50
149 static struct position {
150 	char *name;
151 	unsigned long line;
152 } quotes[MAXRECORD], paren[MAXRECORD];
153 
154 static void record(struct position *, int);
155 static void dump_stack(struct position *, int);
156 
157 static void macro(void);
158 static void initkwds(void);
159 static ndptr inspect(int, char *);
160 static int do_look_ahead(int, const char *);
161 static void reallyoutputstr(const char *);
162 static void reallyputchar(int);
163 
164 static void enlarge_stack(void);
165 
166 int main(int, char *[]);
167 
168 int exit_code = 0;
169 
170 int
171 main(int argc, char *argv[])
172 {
173 	int c;
174 	int n;
175 	char *p;
176 
177 	if (signal(SIGINT, SIG_IGN) != SIG_IGN)
178 		signal(SIGINT, onintr);
179 
180 	init_macros();
181 	initspaces();
182 	STACKMAX = INITSTACKMAX;
183 
184 	mstack = xreallocarray(NULL, STACKMAX, sizeof(stae), NULL);
185 	sstack = xalloc(STACKMAX, NULL);
186 
187 	maxout = 0;
188 	outfile = NULL;
189 	resizedivs(MAXOUT);
190 
191 	while ((c = getopt(argc, argv, "gst:d:D:EU:o:I:P")) != -1)
192 		switch(c) {
193 
194 		case 'D':               /* define something..*/
195 			for (p = optarg; *p; p++)
196 				if (*p == '=')
197 					break;
198 			if (*p)
199 				*p++ = EOS;
200 			dodefine(optarg, p);
201 			break;
202 		case 'E':               /* like GNU m4 1.4.9+ */
203 			if (error_warns == 0)
204 				error_warns = 1;
205 			else
206 				fatal_warns = 1;
207 			break;
208 		case 'I':
209 			addtoincludepath(optarg);
210 			break;
211 		case 'P':
212 			prefix_builtins = 1;
213 			break;
214 		case 'U':               /* undefine...       */
215 			macro_popdef(optarg);
216 			break;
217 		case 'g':
218 			mimic_gnu = 1;
219 			break;
220 		case 'd':
221 			set_trace_flags(optarg);
222 			break;
223 		case 's':
224 			synch_lines = 1;
225 			break;
226 		case 't':
227 			mark_traced(optarg, 1);
228 			break;
229 		case 'o':
230 			trace_file(optarg);
231                         break;
232 		case '?':
233 			usage();
234 		}
235 
236         argc -= optind;
237         argv += optind;
238 
239 	initkwds();
240 	if (mimic_gnu)
241 		setup_builtin("format", FORMATTYPE);
242 
243 	active = stdout;		/* default active output     */
244 	bbase[0] = bufbase;
245         if (!argc) {
246 		sp = -1;		/* stack pointer initialized */
247 		fp = 0;			/* frame pointer initialized */
248 		set_input(infile+0, stdin, "stdin");
249 					/* default input (naturally) */
250 		macro();
251 	} else
252 		for (; argc--; ++argv) {
253 			p = *argv;
254 			if (p[0] == '-' && p[1] == EOS)
255 				set_input(infile, stdin, "stdin");
256 			else if (fopen_trypath(infile, p) == NULL)
257 				err(1, "%s", p);
258 			sp = -1;
259 			fp = 0;
260 			macro();
261 			release_input(infile);
262 		}
263 
264 	if (wrapindex) {
265 		int i;
266 
267 		ilevel = 0;		/* in case m4wrap includes.. */
268 		bufbase = bp = buf;	/* use the entire buffer   */
269 		if (mimic_gnu) {
270 			while (wrapindex != 0) {
271 				for (i = 0; i < wrapindex; i++)
272 					pbstr(m4wraps[i]);
273 				wrapindex =0;
274 				macro();
275 			}
276 		} else {
277 			for (i = 0; i < wrapindex; i++) {
278 				pbstr(m4wraps[i]);
279 				macro();
280 			}
281 		}
282 	}
283 
284 	if (active != stdout)
285 		active = stdout;	/* reset output just in case */
286 	for (n = 1; n < maxout; n++)	/* default wrap-up: undivert */
287 		if (outfile[n] != NULL)
288 			getdiv(n);
289 					/* remove bitbucket if used  */
290 	if (outfile[0] != NULL) {
291 		(void) fclose(outfile[0]);
292 	}
293 
294 	return exit_code;
295 }
296 
297 /*
298  * Look ahead for `token'.
299  * (on input `t == token[0]')
300  * Used for comment and quoting delimiters.
301  * Returns 1 if `token' present; copied to output.
302  *         0 if `token' not found; all characters pushed back
303  */
304 static int
305 do_look_ahead(int t, const char *token)
306 {
307 	int i;
308 
309 	assert((unsigned char)t == (unsigned char)token[0]);
310 
311 	for (i = 1; *++token; i++) {
312 		t = gpbc();
313 		if (t == EOF || (unsigned char)t != (unsigned char)*token) {
314 			pushback(t);
315 			while (--i)
316 				pushback(*--token);
317 			return 0;
318 		}
319 	}
320 	return 1;
321 }
322 
323 #define LOOK_AHEAD(t, token) (t != EOF &&		\
324     (unsigned char)(t)==(unsigned char)(token)[0] &&	\
325     do_look_ahead(t,token))
326 
327 /*
328  * macro - the work horse..
329  */
330 static void
331 macro(void)
332 {
333 	char token[MAXTOK+1];
334 	int t, l;
335 	ndptr p;
336 	int  nlpar;
337 
338 	cycle {
339 		t = gpbc();
340 
341 		if (LOOK_AHEAD(t,lquote)) {	/* strip quotes */
342 			nlpar = 0;
343 			record(quotes, nlpar++);
344 			/*
345 			 * Opening quote: scan forward until matching
346 			 * closing quote has been found.
347 			 */
348 			do {
349 
350 				l = gpbc();
351 				if (LOOK_AHEAD(l,rquote)) {
352 					if (--nlpar > 0)
353 						outputstr(rquote);
354 				} else if (LOOK_AHEAD(l,lquote)) {
355 					record(quotes, nlpar++);
356 					outputstr(lquote);
357 				} else if (l == EOF) {
358 					if (nlpar == 1)
359 						warnx("unclosed quote:");
360 					else
361 						warnx("%d unclosed quotes:", nlpar);
362 					dump_stack(quotes, nlpar);
363 					exit(1);
364 				} else {
365 					if (nlpar > 0) {
366 						if (sp < 0)
367 							reallyputchar(l);
368 						else
369 							CHRSAVE(l);
370 					}
371 				}
372 			}
373 			while (nlpar != 0);
374 		} else if (sp < 0 && LOOK_AHEAD(t, scommt)) {
375 			reallyoutputstr(scommt);
376 
377 			for(;;) {
378 				t = gpbc();
379 				if (LOOK_AHEAD(t, ecommt)) {
380 					reallyoutputstr(ecommt);
381 					break;
382 				}
383 				if (t == EOF)
384 					break;
385 				reallyputchar(t);
386 			}
387 		} else if (t == '_' || isalpha(t)) {
388 			p = inspect(t, token);
389 			if (p != NULL)
390 				pushback(l = gpbc());
391 			if (p == NULL || (l != LPAREN &&
392 			    (macro_getdef(p)->type & NEEDARGS) != 0))
393 				outputstr(token);
394 			else {
395 		/*
396 		 * real thing.. First build a call frame:
397 		 */
398 				pushf(fp);	/* previous call frm */
399 				pushf(macro_getdef(p)->type); /* type of the call  */
400 				pushf(is_traced(p));
401 				pushf(0);	/* parenthesis level */
402 				fp = sp;	/* new frame pointer */
403 		/*
404 		 * now push the string arguments:
405 		 */
406 				pushdef(p);			/* defn string */
407 				pushs1((char *)macro_name(p));	/* macro name  */
408 				pushs(ep);			/* start next..*/
409 
410 				if (l != LPAREN && PARLEV == 0) {
411 				    /* no bracks  */
412 					chrsave(EOS);
413 
414 					if (sp == (int)STACKMAX)
415 						errx(1, "internal stack overflow");
416 					eval((const char **) mstack+fp+1, 2,
417 					    CALTYP, TRACESTATUS);
418 
419 					ep = PREVEP;	/* flush strspace */
420 					sp = PREVSP;	/* previous sp..  */
421 					fp = PREVFP;	/* rewind stack...*/
422 				}
423 			}
424 		} else if (t == EOF) {
425 			if (!mimic_gnu /* you can puke right there */
426 			    && sp > -1 && ilevel <= 0) {
427 				warnx( "unexpected end of input, unclosed parenthesis:");
428 				dump_stack(paren, PARLEV);
429 				exit(1);
430 			}
431 			if (ilevel <= 0)
432 				break;			/* all done thanks.. */
433 			release_input(infile+ilevel--);
434 			emit_synchline();
435 			bufbase = bbase[ilevel];
436 			continue;
437 		} else if (sp < 0) {		/* not in a macro at all */
438 			reallyputchar(t);	/* output directly..	 */
439 		}
440 
441 		else switch(t) {
442 
443 		case LPAREN:
444 			if (PARLEV > 0)
445 				chrsave(t);
446 			while (isspace(l = gpbc())) /* skip blank, tab, nl.. */
447 				if (PARLEV > 0)
448 					chrsave(l);
449 			pushback(l);
450 			record(paren, PARLEV++);
451 			break;
452 
453 		case RPAREN:
454 			if (--PARLEV > 0)
455 				chrsave(t);
456 			else {			/* end of argument list */
457 				chrsave(EOS);
458 
459 				if (sp == (int)STACKMAX)
460 					errx(1, "internal stack overflow");
461 
462 				eval((const char **) mstack+fp+1, sp-fp,
463 				    CALTYP, TRACESTATUS);
464 
465 				ep = PREVEP;	/* flush strspace */
466 				sp = PREVSP;	/* previous sp..  */
467 				fp = PREVFP;	/* rewind stack...*/
468 			}
469 			break;
470 
471 		case COMMA:
472 			if (PARLEV == 1) {
473 				chrsave(EOS);		/* new argument   */
474 				while (isspace(l = gpbc()))
475 					;
476 				pushback(l);
477 				pushs(ep);
478 			} else
479 				chrsave(t);
480 			break;
481 
482 		default:
483 			if (LOOK_AHEAD(t, scommt)) {
484 				char *p;
485 				for (p = scommt; *p; p++)
486 					chrsave(*p);
487 				for(;;) {
488 					t = gpbc();
489 					if (LOOK_AHEAD(t, ecommt)) {
490 						for (p = ecommt; *p; p++)
491 							chrsave(*p);
492 						break;
493 					}
494 					if (t == EOF)
495 					    break;
496 					CHRSAVE(t);
497 				}
498 			} else
499 				CHRSAVE(t);		/* stack the char */
500 			break;
501 		}
502 	}
503 }
504 
505 /*
506  * output string directly, without pushing it for reparses.
507  */
508 void
509 outputstr(const char *s)
510 {
511 	if (sp < 0)
512 		reallyoutputstr(s);
513 	else
514 		while (*s)
515 			CHRSAVE(*s++);
516 }
517 
518 void
519 reallyoutputstr(const char *s)
520 {
521 	if (synch_lines) {
522 		while (*s) {
523 			fputc(*s, active);
524 			if (*s++ == '\n') {
525 				infile[ilevel].synch_lineno++;
526 				if (infile[ilevel].synch_lineno !=
527 				    infile[ilevel].lineno)
528 					do_emit_synchline();
529 			}
530 		}
531 	} else
532 		fputs(s, active);
533 }
534 
535 void
536 reallyputchar(int c)
537 {
538 	putc(c, active);
539 	if (synch_lines && c == '\n') {
540 		infile[ilevel].synch_lineno++;
541 		if (infile[ilevel].synch_lineno != infile[ilevel].lineno)
542 			do_emit_synchline();
543 	}
544 }
545 
546 /*
547  * build an input token..
548  * consider only those starting with _ or A-Za-z.
549  */
550 static ndptr
551 inspect(int c, char *tp)
552 {
553 	char *name = tp;
554 	char *etp = tp+MAXTOK;
555 	ndptr p;
556 
557 	*tp++ = c;
558 
559 	while ((isalnum(c = gpbc()) || c == '_') && tp < etp)
560 		*tp++ = c;
561 	if (c != EOF)
562 		PUSHBACK(c);
563 	*tp = EOS;
564 	/* token is too long, it won't match anything, but it can still
565 	 * be output. */
566 	if (tp == ep) {
567 		outputstr(name);
568 		while (isalnum(c = gpbc()) || c == '_') {
569 			if (sp < 0)
570 				reallyputchar(c);
571 			else
572 				CHRSAVE(c);
573 		}
574 		*name = EOS;
575 		return NULL;
576 	}
577 
578 	p = ohash_find(&macros, ohash_qlookupi(&macros, name, (const char **)&tp));
579 	if (p == NULL)
580 		return NULL;
581 	if (macro_getdef(p) == NULL)
582 		return NULL;
583 	return p;
584 }
585 
586 /*
587  * initkwds - initialise m4 keywords as fast as possible.
588  * This very similar to install, but without certain overheads,
589  * such as calling lookup. Malloc is not used for storing the
590  * keyword strings, since we simply use the static pointers
591  * within keywrds block.
592  */
593 static void
594 initkwds(void)
595 {
596 	unsigned int type;
597 	int i;
598 
599 	for (i = 0; i < (int)MAXKEYS; i++) {
600 		type = keywrds[i].ktyp & TYPEMASK;
601 		if ((keywrds[i].ktyp & NOARGS) == 0)
602 			type |= NEEDARGS;
603 		setup_builtin(keywrds[i].knam, type);
604 	}
605 }
606 
607 static void
608 record(struct position *t, int lev)
609 {
610 	if (lev < MAXRECORD) {
611 		t[lev].name = CURRENT_NAME;
612 		t[lev].line = CURRENT_LINE;
613 	}
614 }
615 
616 static void
617 dump_stack(struct position *t, int lev)
618 {
619 	int i;
620 
621 	for (i = 0; i < lev; i++) {
622 		if (i == MAXRECORD) {
623 			fprintf(stderr, "   ...\n");
624 			break;
625 		}
626 		fprintf(stderr, "   %s at line %lu\n",
627 			t[i].name, t[i].line);
628 	}
629 }
630 
631 
632 static void
633 enlarge_stack(void)
634 {
635 	STACKMAX += STACKMAX/2;
636 	mstack = xreallocarray(mstack, STACKMAX, sizeof(stae),
637 	    "Evaluation stack overflow (%lu)",
638 	    (unsigned long)STACKMAX);
639 	sstack = xrealloc(sstack, STACKMAX,
640 	    "Evaluation stack overflow (%lu)",
641 	    (unsigned long)STACKMAX);
642 }
643