xref: /netbsd-src/usr.bin/config/scan.l (revision c8da0e5fefd3800856b306200a18b2315c7fbb9f)
1 %{
2 /*	$NetBSD: scan.l,v 1.12 2009/04/11 12:41:10 lukem Exp $	*/
3 
4 /*
5  * Copyright (c) 1992, 1993
6  *	The Regents of the University of California.  All rights reserved.
7  *
8  * This software was developed by the Computer Systems Engineering group
9  * at Lawrence Berkeley Laboratory under DARPA contract BG 91-66 and
10  * contributed to Berkeley.
11  *
12  * All advertising materials mentioning features or use of this software
13  * must display the following acknowledgement:
14  *	This product includes software developed by the University of
15  *	California, Lawrence Berkeley Laboratories.
16  *
17  * Redistribution and use in source and binary forms, with or without
18  * modification, are permitted provided that the following conditions
19  * are met:
20  * 1. Redistributions of source code must retain the above copyright
21  *    notice, this list of conditions and the following disclaimer.
22  * 2. Redistributions in binary form must reproduce the above copyright
23  *    notice, this list of conditions and the following disclaimer in the
24  *    documentation and/or other materials provided with the distribution.
25  * 3. Neither the name of the University nor the names of its contributors
26  *    may be used to endorse or promote products derived from this software
27  *    without specific prior written permission.
28  *
29  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
30  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
31  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
32  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
33  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
34  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
35  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
36  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
37  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
38  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
39  * SUCH DAMAGE.
40  *
41  *	from: @(#)scan.l	8.1 (Berkeley) 6/6/93
42  */
43 
44 #include <sys/param.h>
45 #include <errno.h>
46 #include <libgen.h>
47 #include <stdio.h>
48 #include <stdlib.h>
49 #include <string.h>
50 #include <unistd.h>
51 #include <stddef.h>
52 #include <ctype.h>
53 #include <util.h>
54 #undef ECHO
55 #include "defs.h"
56 #include "gram.h"
57 
58 int	yyline;
59 const char *yyfile;
60 const char *lastfile;
61 char curinclpath[PATH_MAX];
62 int ifdefstate = -1;
63 int st;
64 #define IDS_PARENT_DISABLED \
65     ((ifdefstate > 6) && ((((ifdefstate/6)-1) & 1) == 1))
66 #define IDS_MAX_DEPTH		362797056 /* 6^11 */
67 /* States for ifdefstate:
68 
69   0  -> matched ifdef
70   1  -> unmatched ifdef
71   2  -> matched elifdef
72   3  -> unmatched elifdef
73   4  -> matched else
74   5  -> unmatched else
75 
76   Upon "ifdef", add one and multiply by 6.
77   Upon "endif", divide by 6, remove 1.
78 
79   ifdef -> MATCH => continue
80            MISMATCH => set to 1
81   elifdef -> if (!1) -> MISMATCH
82              MATCH => set to 2
83              MISMATCH => if (2 || 3) set to 3, else set to 1
84   else -> if (1) -> MATCH
85           MATCH => set to 4
86           MISMATCH => set to 5
87 
88   in each case, if parent & 1 == 1, MISMATCH
89 */
90 
91 /*
92  * Data for returning to previous files from include files.
93  */
94 struct incl {
95 	struct	incl *in_prev;	/* previous includes in effect, if any */
96 	YY_BUFFER_STATE in_buf;	/* previous lex state */
97 	const char *in_fname;	/* previous file name */
98 	int	in_lineno;	/* previous line number */
99 	int	in_ateof;	/* token to insert at EOF */
100 	int	in_interesting;	/* previous value for "interesting" */
101 	int	in_ifdefstate;	/* conditional level */
102 };
103 static struct incl *incl;
104 static int endinclude(void);
105 static int getincludepath(void);
106 static int getcurifdef(void);
107 
108 #define	yywrap() 1
109 
110 %}
111 
112 PATH	[A-Za-z_0-9]*[./][-A-Za-z_0-9./]*
113 QCHARS	([^"\n]|\\\")+
114 WORD	[A-Za-z_][-A-Za-z_0-9]*
115 FILENAME	({PATH}|\"{QCHARS}\")
116 RESTOFLINE	[ \t]*(#[^\n]*)?\n
117 
118 %x	IGNORED
119 
120 %%
121 		/* Local variables for yylex() */
122 		int tok;
123 
124 and		return AND;
125 at		return AT;
126 attach		return ATTACH;
127 block		return BLOCK;
128 build		return BUILD;
129 char		return CHAR;
130 compile-with	return COMPILE_WITH;
131 config		return CONFIG;
132 deffs		return DEFFS;
133 define		return DEFINE;
134 defflag		return DEFFLAG;
135 defopt		return DEFOPT;
136 defparam	return DEFPARAM;
137 defpseudo	return DEFPSEUDO;
138 defpseudodev	return DEFPSEUDODEV;
139 devclass	return DEVCLASS;
140 device		return DEVICE;
141 device-major	return DEVICE_MAJOR;
142 dumps		return DUMPS;
143 file		return XFILE;
144 file-system	return FILE_SYSTEM;
145 flags		return FLAGS;
146 ident		return IDENT;
147 machine		return XMACHINE;
148 major		return MAJOR;
149 makeoptions	return MAKEOPTIONS;
150 maxpartitions	return MAXPARTITIONS;
151 maxusers	return MAXUSERS;
152 minor		return MINOR;
153 needs-count	return NEEDS_COUNT;
154 needs-flag	return NEEDS_FLAG;
155 no		return NO;
156 object		return XOBJECT;
157 obsolete	return OBSOLETE;
158 on		return ON;
159 options		return OPTIONS;
160 prefix		return PREFIX;
161 pseudo-device	return PSEUDO_DEVICE;
162 root		return ROOT;
163 source		return SOURCE;
164 type		return TYPE;
165 version 	return VERSION;
166 with		return WITH;
167 
168 \+=		return PLUSEQ;
169 :=		return COLONEQ;
170 
171 <*>ifdef[ \t]+{WORD}{RESTOFLINE} {
172 		ifdefstate = (ifdefstate + 1) * 6;
173 		if (ifdefstate >= IDS_MAX_DEPTH) {
174 			yyerror("too many levels of conditional");
175 		}
176 		if (!IDS_PARENT_DISABLED && getcurifdef()) {
177 			BEGIN(INITIAL);
178 		} else {
179 			ifdefstate++;
180 			BEGIN(IGNORED);
181 		}
182 		yyline++;
183 	}
184 
185 <*>ifndef[ \t]+{WORD}{RESTOFLINE} {
186 		ifdefstate = (ifdefstate + 1) * 6;
187 		if (ifdefstate >= IDS_MAX_DEPTH) {
188 			yyerror("too many levels of conditional");
189 		}
190 		if (!IDS_PARENT_DISABLED && !getcurifdef()) {
191 			BEGIN(INITIAL);
192 		} else {
193 			ifdefstate++;
194 			BEGIN(IGNORED);
195 		}
196 		yyline++;
197 	}
198 
199 
200 <*>elifdef[ \t]+{WORD}{RESTOFLINE} {
201 		st = ifdefstate % 6;
202 		if (ifdefstate < 0 || st > 3) {
203 			yyerror("mismatched elifdef");
204 		}
205 		if (IDS_PARENT_DISABLED ||
206 		    st != 1 || !getcurifdef()) {
207 			if (st == 2 || st == 3) {
208 				ifdefstate += 3 - st;
209 			} else {
210 				ifdefstate += 1 - st;
211 			}
212 			BEGIN(IGNORED);
213 		} else {
214 			ifdefstate++;
215 			BEGIN(INITIAL);
216 		}
217 		yyline++;
218 	}
219 
220 <*>elifndef[ \t]+{WORD}{RESTOFLINE} {
221 		st = ifdefstate % 6;
222 		if (ifdefstate < 0 || st > 3) {
223 			yyerror("mismatched elifndef");
224 		}
225 		if (IDS_PARENT_DISABLED ||
226 		    st != 1 || getcurifdef()) {
227 			if (st == 2 || st == 3) {
228 				ifdefstate += 3 - st;
229 			} else {
230 				ifdefstate += 1 - st;
231 			}
232 			BEGIN(IGNORED);
233 		} else {
234 			ifdefstate++;
235 			BEGIN(INITIAL);
236 		}
237 		yyline++;
238 	}
239 
240 <*>else{RESTOFLINE} {
241 		st = ifdefstate % 6;
242 		if (ifdefstate < 0 || st > 3) {
243 			yyerror("mismatched else");
244 		}
245 		if (!IDS_PARENT_DISABLED && (st == 1)) {
246 			ifdefstate += 3;
247 			BEGIN(INITIAL);
248 		} else {
249 			ifdefstate += 5 - st;
250 			BEGIN(IGNORED);
251 		}
252 		yyline++;
253 	}
254 
255 <*>endif{RESTOFLINE} {
256 		if (ifdefstate < 0) {
257 			yyerror("mismatched endif");
258 		}
259 		if (!IDS_PARENT_DISABLED) {
260 			BEGIN(INITIAL);
261 		}
262 		ifdefstate = (ifdefstate/6) - 1;
263 		yyline++;
264 	}
265 
266 <IGNORED>\n		{
267 		yyline++;
268 	}
269 
270 <IGNORED>.	/* ignore */
271 
272 include[ \t]+{FILENAME}{RESTOFLINE}	{
273 		yyline++;
274 		if (getincludepath()) {
275 			include(curinclpath, 0, 0, 1);
276 		} else {
277 			yyerror("bad include path-name");
278 		}
279 	}
280 
281 cinclude[ \t]+{FILENAME}{RESTOFLINE}	{
282 		yyline++;
283 		if (getincludepath()) {
284 			include(curinclpath, 0, 1, 1);
285 		} else {
286 			yyerror("bad cinclude path-name");
287 		}
288 	}
289 
290 package[ \t]+{FILENAME}{RESTOFLINE}	{
291 		yyline++;
292 		if (!oktopackage) {
293 			yyerror("package not allowed here");
294 		} else if (getincludepath()) {
295 			package(curinclpath);
296 		} else {
297 			yyerror("bad package path-name");
298 		}
299 	}
300 
301 {PATH}	{
302 		yylval.str = intern(yytext);
303 		return PATHNAME;
304 	}
305 
306 {WORD}	{
307 		yylval.str = intern(yytext);
308 		return WORD;
309 	}
310 
311 \"\" {
312 		yylval.str = intern("");
313 		return EMPTYSTRING;
314 	}
315 
316 \"{QCHARS}	{
317 		tok = input();  /* eat closing quote */
318 		if (tok != '"') {
319 			cfgerror("closing quote missing\n");
320 			unput(tok);
321 		}
322 		yylval.str = intern(yytext + 1);
323 		return QSTRING;
324 	}
325 0[0-7]*	{
326 		yylval.num.fmt = 8;
327 		yylval.num.val = strtoll(yytext, NULL, 8);
328 		return NUMBER;
329 	}
330 0[xX][0-9a-fA-F]+ {
331 		yylval.num.fmt = 16;
332 		yylval.num.val = strtoull(yytext + 2, NULL, 16);
333 		return NUMBER;
334 	}
335 [1-9][0-9]* {
336 		yylval.num.fmt = 10;
337 		yylval.num.val = strtoll(yytext, NULL, 10);
338 		return NUMBER;
339 	}
340 \n[ \t] {
341 		/*
342 		 * Note: newline followed by whitespace is always a
343 		 * continuation of the previous line, so do NOT
344 		 * return a token in this case.
345 		 */
346 		yyline++;
347 	}
348 \n	{
349 		yyline++;
350 		return '\n';
351 	}
352 \00	{
353 		/* Detect NUL characters in the config file and
354 		 * error out.
355 		 */
356 		cfgerror("NUL character detected at line %i\n", yyline);
357 	}
358 #.*	{ /* ignored (comment) */; }
359 [ \t]+	{ /* ignored (white space) */; }
360 .	{ return yytext[0]; }
361 <*><<EOF>> {
362 		if (ifdefstate > (incl == NULL ? -1 : incl->in_ifdefstate)) {
363 			yyerror("reached EOF while looking for endif");
364 		}
365 		if (incl == NULL)
366 			return YY_NULL;
367 		tok = endinclude();
368 		if (tok)
369 			return tok;
370 		/* otherwise continue scanning */
371 	}
372 
373 %%
374 
375 int interesting = 1;
376 
377 static int
378 curdir_push(const char *fname)
379 {
380 	struct prefix *pf;
381 	char *p, *d, *f;
382 
383 	/* Set up the initial "current directory" for include directives. */
384 	d = dirname(f = estrdup(fname));
385 	if (*d == '/')
386 		p = estrdup(d);
387 	else {
388 		char *cwd, buf[PATH_MAX];
389 
390 		if ((cwd = getcwd(buf, sizeof(buf))) == NULL) {
391 			free(f);
392 			return (-1);
393 		}
394 		p = emalloc(strlen(cwd) + strlen(d) + 2);
395 		sprintf(p, "%s/%s", cwd, d);
396 	}
397 	free(f);
398 	pf = ecalloc(1, sizeof(*pf));
399 	pf->pf_prefix = p;
400 	SLIST_INSERT_HEAD(&curdirs, pf, pf_next);
401 
402 	return (0);
403 }
404 
405 static void
406 curdir_pop(void)
407 {
408 	struct prefix *pf;
409 
410 	pf = SLIST_FIRST(&curdirs);
411 	SLIST_REMOVE_HEAD(&curdirs, pf_next);
412 	if (SLIST_EMPTY(&curdirs))
413 		panic("curdirs is empty");
414 	/* LINTED cast away const (pf_prefix is malloc'd for curdirs) */
415 	free((void *)__UNCONST(pf->pf_prefix));
416 	free(pf);
417 }
418 
419 /*
420  * Open the "main" file (conffile).
421  */
422 int
423 firstfile(const char *fname)
424 {
425 
426 #if defined(__NetBSD__)
427 	if ((yyin = fopen(fname, "rf")) == NULL)
428 #else
429 	if ((yyin = fopen(fname, "r")) == NULL)
430 #endif
431 		return (-1);
432 
433 	if (curdir_push(fname) == -1)
434 		return (-1);
435 
436 	yyfile = conffile = fname;
437 	yyline = 1;
438 	return (0);
439 }
440 
441 /*
442  * Add a "package" to the configuration.  This is essentially
443  * syntactic sugar around the sequence:
444  *
445  *	prefix ../some/directory
446  *	include "files.package"
447  *	prefix
448  */
449 void
450 package(const char *fname)
451 {
452 	char *fname1 = estrdup(fname);
453 	char *fname2 = estrdup(fname);
454 	char *dir = dirname(fname1);
455 	char *file = basename(fname2);
456 
457 	/*
458 	 * Push the prefix on to the prefix stack and process the include
459 	 * file.  When we reach the end of the include file, inserting
460 	 * the PREFIX token into the input stream will pop the prefix off
461 	 * of the prefix stack.
462 	 */
463 	prefix_push(dir);
464 	(void) include(file, PREFIX, 0, 1);
465 
466 	free(fname1);
467 	free(fname2);
468 }
469 
470 /*
471  * Open the named file for inclusion at the current point.  Returns 0 on
472  * success (file opened and previous state pushed), nonzero on failure
473  * (fopen failed, complaint made).  The `ateof' parameter controls the
474  * token to be inserted at the end of the include file (i.e. ENDFILE).
475  * If ateof == 0 then nothing is inserted.
476  */
477 int
478 include(const char *fname, int ateof, int conditional, int direct)
479 {
480 	FILE *fp;
481 	struct incl *in;
482 	char *s;
483 	static int havedirs;
484 	extern int vflag;
485 
486 	if (havedirs == 0) {
487 		havedirs = 1;
488 		setupdirs();
489 	}
490 
491 	if (fname[0] == '/')
492 		s = estrdup(fname);
493 	else if (fname[0] == '.' && fname[1] == '/') {
494 		struct prefix *pf = SLIST_FIRST(&curdirs);
495 		s = emalloc(strlen(pf->pf_prefix) + strlen(fname));
496 		sprintf(s, "%s/%s", pf->pf_prefix, fname + 2);
497 	} else
498 		s = sourcepath(fname);
499 	if ((fp = fopen(s, "r")) == NULL) {
500 		if (conditional == 0)
501 			cfgerror("cannot open %s for reading: %s\n", s,
502 			    strerror(errno));
503 		else if (vflag)
504 			cfgwarn("cannot open conditional include file %s: %s",
505 			     s, strerror(errno));
506 		free(s);
507 		return (-1);
508 	}
509 	if (curdir_push(s) == -1) {
510 		cfgerror("cannot record current working directory for %s\n", s);
511 		fclose(fp);
512 		free(s);
513 		return (-1);
514 	}
515 	in = ecalloc(1, sizeof *in);
516 	in->in_prev = incl;
517 	in->in_buf = YY_CURRENT_BUFFER;
518 	in->in_fname = yyfile;
519 	in->in_lineno = yyline;
520 	in->in_ateof = ateof;
521 	in->in_interesting = interesting;
522 	in->in_ifdefstate = ifdefstate;
523 	interesting = direct & interesting;
524 	if (interesting)
525 		logconfig_include(fp, fname);
526 	incl = in;
527 	yy_switch_to_buffer(yy_create_buffer(fp, YY_BUF_SIZE));
528 	yyfile = intern(s);
529 	yyline = 1;
530 	free(s);
531 	return (0);
532 }
533 
534 /*
535  * Extract the pathname from a include/cinclude/package into curinclpath
536  */
537 static int
538 getincludepath()
539 {
540 	const char *p = yytext;
541 	ptrdiff_t len;
542 	const char *e;
543 
544 	while (*p && isascii((unsigned int)*p) && !isspace((unsigned int)*p))
545 		p++;
546 	while (*p && isascii((unsigned int)*p) && isspace((unsigned int)*p))
547 		p++;
548 	if (!*p)
549 		return 0;
550 	if (*p == '"') {
551 		p++;
552 		e = strchr(p, '"');
553 		if (!e) return 0;
554 	} else {
555 		e = p;
556 		while (*e && isascii((unsigned int)*e)
557 		    && !isspace((unsigned int)*e))
558 			e++;
559 	}
560 
561 	len = e-p;
562 	if (len > (ptrdiff_t)sizeof(curinclpath)-1)
563 		len = sizeof(curinclpath)-1;
564 	strncpy(curinclpath, p, sizeof(curinclpath));
565 	curinclpath[len] = '\0';
566 
567 	return 1;
568 }
569 
570 /*
571  * Terminate the most recent inclusion.
572  */
573 static int
574 endinclude(void)
575 {
576 	struct incl *in;
577 	int ateof;
578 
579 	curdir_pop();
580 	if ((in = incl) == NULL)
581 		panic("endinclude");
582 	incl = in->in_prev;
583 	lastfile = yyfile;
584 	yy_delete_buffer(YY_CURRENT_BUFFER);
585 	(void)fclose(yyin);
586 	yy_switch_to_buffer(in->in_buf);
587 	yyfile = in->in_fname;
588 	yyline = in->in_lineno;
589 	ateof  = in->in_ateof;
590 	interesting = in->in_interesting;
591 	free(in);
592 
593 	return (ateof);
594 }
595 
596 /*
597  * Return the current line number.  If yacc has looked ahead and caused
598  * us to consume a newline, we have to subtract one.  yychar is yacc's
599  * token lookahead, so we can tell.
600  */
601 int
602 currentline(void)
603 {
604 	extern int yychar;
605 
606 	return (yyline - (yychar == '\n'));
607 }
608 
609 static int
610 getcurifdef(void)
611 {
612 	char *p = yytext, *q;
613 
614 	while (*p && isascii((unsigned int)*p) && !isspace((unsigned int)*p))
615 		p++;
616 	while (*p && isascii((unsigned int)*p) && isspace((unsigned int)*p))
617 		p++;
618 	q = p;
619 	while (*q && isascii((unsigned int)*q) && !isspace((unsigned int)*q))
620 		q++;
621 	*q = '\0';
622 
623 	return ht_lookup(attrtab, intern(p)) != NULL;
624 }
625