xref: /freebsd/bin/sh/expand.c (revision 6780ab54325a71e7e70112b11657973edde8655e)
1 /*-
2  * Copyright (c) 1991, 1993
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * This code is derived from software contributed to Berkeley by
6  * Kenneth Almquist.
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  * 3. All advertising materials mentioning features or use of this software
17  *    must display the following acknowledgement:
18  *	This product includes software developed by the University of
19  *	California, Berkeley and its contributors.
20  * 4. Neither the name of the University nor the names of its contributors
21  *    may be used to endorse or promote products derived from this software
22  *    without specific prior written permission.
23  *
24  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34  * SUCH DAMAGE.
35  */
36 
37 #ifndef lint
38 #if 0
39 static char sccsid[] = "@(#)expand.c	8.5 (Berkeley) 5/15/95";
40 #endif
41 #endif /* not lint */
42 #include <sys/cdefs.h>
43 __FBSDID("$FreeBSD$");
44 
45 #include <sys/types.h>
46 #include <sys/time.h>
47 #include <sys/stat.h>
48 #include <errno.h>
49 #include <dirent.h>
50 #include <unistd.h>
51 #include <pwd.h>
52 #include <stdlib.h>
53 #include <limits.h>
54 #include <stdio.h>
55 #include <string.h>
56 
57 /*
58  * Routines to expand arguments to commands.  We have to deal with
59  * backquotes, shell variables, and file metacharacters.
60  */
61 
62 #include "shell.h"
63 #include "main.h"
64 #include "nodes.h"
65 #include "eval.h"
66 #include "expand.h"
67 #include "syntax.h"
68 #include "parser.h"
69 #include "jobs.h"
70 #include "options.h"
71 #include "var.h"
72 #include "input.h"
73 #include "output.h"
74 #include "memalloc.h"
75 #include "error.h"
76 #include "mystring.h"
77 #include "arith.h"
78 #include "show.h"
79 
80 /*
81  * Structure specifying which parts of the string should be searched
82  * for IFS characters.
83  */
84 
85 struct ifsregion {
86 	struct ifsregion *next;	/* next region in list */
87 	int begoff;		/* offset of start of region */
88 	int endoff;		/* offset of end of region */
89 	int nulonly;		/* search for nul bytes only */
90 };
91 
92 
93 char *expdest;			/* output of current string */
94 struct nodelist *argbackq;	/* list of back quote expressions */
95 struct ifsregion ifsfirst;	/* first struct in list of ifs regions */
96 struct ifsregion *ifslastp;	/* last struct in list */
97 struct arglist exparg;		/* holds expanded arg list */
98 
99 STATIC void argstr(char *, int);
100 STATIC char *exptilde(char *, int);
101 STATIC void expbackq(union node *, int, int);
102 STATIC int subevalvar(char *, char *, int, int, int, int);
103 STATIC char *evalvar(char *, int);
104 STATIC int varisset(char *, int);
105 STATIC void varvalue(char *, int, int);
106 STATIC void recordregion(int, int, int);
107 STATIC void removerecordregions(int);
108 STATIC void ifsbreakup(char *, struct arglist *);
109 STATIC void expandmeta(struct strlist *, int);
110 STATIC void expmeta(char *, char *);
111 STATIC void addfname(char *);
112 STATIC struct strlist *expsort(struct strlist *);
113 STATIC struct strlist *msort(struct strlist *, int);
114 STATIC int pmatch(char *, char *, int);
115 STATIC char *cvtnum(int, char *);
116 STATIC int collate_range_cmp(int, int);
117 
118 STATIC int
119 collate_range_cmp (int c1, int c2)
120 {
121 	static char s1[2], s2[2];
122 	int ret;
123 
124 	c1 &= UCHAR_MAX;
125 	c2 &= UCHAR_MAX;
126 	if (c1 == c2)
127 		return (0);
128 	s1[0] = c1;
129 	s2[0] = c2;
130 	if ((ret = strcoll(s1, s2)) != 0)
131 		return (ret);
132 	return (c1 - c2);
133 }
134 
135 /*
136  * Expand shell variables and backquotes inside a here document.
137  *	union node *arg		the document
138  *	int fd;			where to write the expanded version
139  */
140 
141 void
142 expandhere(union node *arg, int fd)
143 {
144 	herefd = fd;
145 	expandarg(arg, (struct arglist *)NULL, 0);
146 	xwrite(fd, stackblock(), expdest - stackblock());
147 }
148 
149 
150 /*
151  * Perform variable substitution and command substitution on an argument,
152  * placing the resulting list of arguments in arglist.  If EXP_FULL is true,
153  * perform splitting and file name expansion.  When arglist is NULL, perform
154  * here document expansion.
155  */
156 
157 void
158 expandarg(union node *arg, struct arglist *arglist, int flag)
159 {
160 	struct strlist *sp;
161 	char *p;
162 
163 	argbackq = arg->narg.backquote;
164 	STARTSTACKSTR(expdest);
165 	ifsfirst.next = NULL;
166 	ifslastp = NULL;
167 	argstr(arg->narg.text, flag);
168 	if (arglist == NULL) {
169 		return;			/* here document expanded */
170 	}
171 	STPUTC('\0', expdest);
172 	p = grabstackstr(expdest);
173 	exparg.lastp = &exparg.list;
174 	/*
175 	 * TODO - EXP_REDIR
176 	 */
177 	if (flag & EXP_FULL) {
178 		ifsbreakup(p, &exparg);
179 		*exparg.lastp = NULL;
180 		exparg.lastp = &exparg.list;
181 		expandmeta(exparg.list, flag);
182 	} else {
183 		if (flag & EXP_REDIR) /*XXX - for now, just remove escapes */
184 			rmescapes(p);
185 		sp = (struct strlist *)stalloc(sizeof (struct strlist));
186 		sp->text = p;
187 		*exparg.lastp = sp;
188 		exparg.lastp = &sp->next;
189 	}
190 	while (ifsfirst.next != NULL) {
191 		struct ifsregion *ifsp;
192 		INTOFF;
193 		ifsp = ifsfirst.next->next;
194 		ckfree(ifsfirst.next);
195 		ifsfirst.next = ifsp;
196 		INTON;
197 	}
198 	*exparg.lastp = NULL;
199 	if (exparg.list) {
200 		*arglist->lastp = exparg.list;
201 		arglist->lastp = exparg.lastp;
202 	}
203 }
204 
205 
206 
207 /*
208  * Perform variable and command substitution.  If EXP_FULL is set, output CTLESC
209  * characters to allow for further processing.  Otherwise treat
210  * $@ like $* since no splitting will be performed.
211  */
212 
213 STATIC void
214 argstr(char *p, int flag)
215 {
216 	char c;
217 	int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR);	/* do CTLESC */
218 	int firsteq = 1;
219 
220 	if (*p == '~' && (flag & (EXP_TILDE | EXP_VARTILDE)))
221 		p = exptilde(p, flag);
222 	for (;;) {
223 		switch (c = *p++) {
224 		case '\0':
225 		case CTLENDVAR: /* ??? */
226 			goto breakloop;
227 		case CTLQUOTEMARK:
228 			/* "$@" syntax adherence hack */
229 			if (p[0] == CTLVAR && p[2] == '@' && p[3] == '=')
230 				break;
231 			if ((flag & EXP_FULL) != 0)
232 				STPUTC(c, expdest);
233 			break;
234 		case CTLESC:
235 			if (quotes)
236 				STPUTC(c, expdest);
237 			c = *p++;
238 			STPUTC(c, expdest);
239 			break;
240 		case CTLVAR:
241 			p = evalvar(p, flag);
242 			break;
243 		case CTLBACKQ:
244 		case CTLBACKQ|CTLQUOTE:
245 			expbackq(argbackq->n, c & CTLQUOTE, flag);
246 			argbackq = argbackq->next;
247 			break;
248 		case CTLENDARI:
249 			expari(flag);
250 			break;
251 		case ':':
252 		case '=':
253 			/*
254 			 * sort of a hack - expand tildes in variable
255 			 * assignments (after the first '=' and after ':'s).
256 			 */
257 			STPUTC(c, expdest);
258 			if (flag & EXP_VARTILDE && *p == '~') {
259 				if (c == '=') {
260 					if (firsteq)
261 						firsteq = 0;
262 					else
263 						break;
264 				}
265 				p = exptilde(p, flag);
266 			}
267 			break;
268 		default:
269 			STPUTC(c, expdest);
270 		}
271 	}
272 breakloop:;
273 }
274 
275 STATIC char *
276 exptilde(char *p, int flag)
277 {
278 	char c, *startp = p;
279 	struct passwd *pw;
280 	char *home;
281 	int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR);
282 
283 	while ((c = *p) != '\0') {
284 		switch(c) {
285 		case CTLESC:
286 			return (startp);
287 		case CTLQUOTEMARK:
288 			return (startp);
289 		case ':':
290 			if (flag & EXP_VARTILDE)
291 				goto done;
292 			break;
293 		case '/':
294 			goto done;
295 		}
296 		p++;
297 	}
298 done:
299 	*p = '\0';
300 	if (*(startp+1) == '\0') {
301 		if ((home = lookupvar("HOME")) == NULL)
302 			goto lose;
303 	} else {
304 		if ((pw = getpwnam(startp+1)) == NULL)
305 			goto lose;
306 		home = pw->pw_dir;
307 	}
308 	if (*home == '\0')
309 		goto lose;
310 	*p = c;
311 	while ((c = *home++) != '\0') {
312 		if (quotes && SQSYNTAX[(int)c] == CCTL)
313 			STPUTC(CTLESC, expdest);
314 		STPUTC(c, expdest);
315 	}
316 	return (p);
317 lose:
318 	*p = c;
319 	return (startp);
320 }
321 
322 
323 STATIC void
324 removerecordregions(int endoff)
325 {
326 	if (ifslastp == NULL)
327 		return;
328 
329 	if (ifsfirst.endoff > endoff) {
330 		while (ifsfirst.next != NULL) {
331 			struct ifsregion *ifsp;
332 			INTOFF;
333 			ifsp = ifsfirst.next->next;
334 			ckfree(ifsfirst.next);
335 			ifsfirst.next = ifsp;
336 			INTON;
337 		}
338 		if (ifsfirst.begoff > endoff)
339 			ifslastp = NULL;
340 		else {
341 			ifslastp = &ifsfirst;
342 			ifsfirst.endoff = endoff;
343 		}
344 		return;
345 	}
346 
347 	ifslastp = &ifsfirst;
348 	while (ifslastp->next && ifslastp->next->begoff < endoff)
349 		ifslastp=ifslastp->next;
350 	while (ifslastp->next != NULL) {
351 		struct ifsregion *ifsp;
352 		INTOFF;
353 		ifsp = ifslastp->next->next;
354 		ckfree(ifslastp->next);
355 		ifslastp->next = ifsp;
356 		INTON;
357 	}
358 	if (ifslastp->endoff > endoff)
359 		ifslastp->endoff = endoff;
360 }
361 
362 /*
363  * Expand arithmetic expression.  Backup to start of expression,
364  * evaluate, place result in (backed up) result, adjust string position.
365  */
366 void
367 expari(int flag)
368 {
369 	char *p, *start;
370 	int result;
371 	int begoff;
372 	int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR);
373 	int quoted;
374 
375 
376 	/*
377 	 * This routine is slightly over-complicated for
378 	 * efficiency.  First we make sure there is
379 	 * enough space for the result, which may be bigger
380 	 * than the expression if we add exponentiation.  Next we
381 	 * scan backwards looking for the start of arithmetic.  If the
382 	 * next previous character is a CTLESC character, then we
383 	 * have to rescan starting from the beginning since CTLESC
384 	 * characters have to be processed left to right.
385 	 */
386 #if INT_MAX / 1000000000 >= 10 || INT_MIN / 1000000000 <= -10
387 #error "integers with more than 10 digits are not supported"
388 #endif
389 	CHECKSTRSPACE(12 - 2, expdest);
390 	USTPUTC('\0', expdest);
391 	start = stackblock();
392 	p = expdest - 2;
393 	while (p >= start && *p != CTLARI)
394 		--p;
395 	if (p < start || *p != CTLARI)
396 		error("missing CTLARI (shouldn't happen)");
397 	if (p > start && *(p - 1) == CTLESC)
398 		for (p = start; *p != CTLARI; p++)
399 			if (*p == CTLESC)
400 				p++;
401 
402 	if (p[1] == '"')
403 		quoted=1;
404 	else
405 		quoted=0;
406 	begoff = p - start;
407 	removerecordregions(begoff);
408 	if (quotes)
409 		rmescapes(p+2);
410 	result = arith(p+2);
411 	fmtstr(p, 12, "%d", result);
412 	while (*p++)
413 		;
414 	if (quoted == 0)
415 		recordregion(begoff, p - 1 - start, 0);
416 	result = expdest - p + 1;
417 	STADJUST(-result, expdest);
418 }
419 
420 
421 /*
422  * Expand stuff in backwards quotes.
423  */
424 
425 STATIC void
426 expbackq(union node *cmd, int quoted, int flag)
427 {
428 	struct backcmd in;
429 	int i;
430 	char buf[128];
431 	char *p;
432 	char *dest = expdest;
433 	struct ifsregion saveifs, *savelastp;
434 	struct nodelist *saveargbackq;
435 	char lastc;
436 	int startloc = dest - stackblock();
437 	char const *syntax = quoted? DQSYNTAX : BASESYNTAX;
438 	int saveherefd;
439 	int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR);
440 
441 	INTOFF;
442 	saveifs = ifsfirst;
443 	savelastp = ifslastp;
444 	saveargbackq = argbackq;
445 	saveherefd = herefd;
446 	herefd = -1;
447 	p = grabstackstr(dest);
448 	evalbackcmd(cmd, &in);
449 	ungrabstackstr(p, dest);
450 	ifsfirst = saveifs;
451 	ifslastp = savelastp;
452 	argbackq = saveargbackq;
453 	herefd = saveherefd;
454 
455 	p = in.buf;
456 	lastc = '\0';
457 	for (;;) {
458 		if (--in.nleft < 0) {
459 			if (in.fd < 0)
460 				break;
461 			while ((i = read(in.fd, buf, sizeof buf)) < 0 && errno == EINTR);
462 			TRACE(("expbackq: read returns %d\n", i));
463 			if (i <= 0)
464 				break;
465 			p = buf;
466 			in.nleft = i - 1;
467 		}
468 		lastc = *p++;
469 		if (lastc != '\0') {
470 			if (quotes && syntax[(int)lastc] == CCTL)
471 				STPUTC(CTLESC, dest);
472 			STPUTC(lastc, dest);
473 		}
474 	}
475 
476 	/* Eat all trailing newlines */
477 	for (p--; lastc == '\n'; lastc = *--p)
478 		STUNPUTC(dest);
479 
480 	if (in.fd >= 0)
481 		close(in.fd);
482 	if (in.buf)
483 		ckfree(in.buf);
484 	if (in.jp)
485 		exitstatus = waitforjob(in.jp, (int *)NULL);
486 	if (quoted == 0)
487 		recordregion(startloc, dest - stackblock(), 0);
488 	TRACE(("evalbackq: size=%d: \"%.*s\"\n",
489 		(dest - stackblock()) - startloc,
490 		(dest - stackblock()) - startloc,
491 		stackblock() + startloc));
492 	expdest = dest;
493 	INTON;
494 }
495 
496 
497 
498 STATIC int
499 subevalvar(char *p, char *str, int strloc, int subtype, int startloc,
500   int varflags)
501 {
502 	char *startp;
503 	char *loc = NULL;
504 	char *q;
505 	int c = 0;
506 	int saveherefd = herefd;
507 	struct nodelist *saveargbackq = argbackq;
508 	int amount;
509 
510 	herefd = -1;
511 	argstr(p, 0);
512 	STACKSTRNUL(expdest);
513 	herefd = saveherefd;
514 	argbackq = saveargbackq;
515 	startp = stackblock() + startloc;
516 	if (str == NULL)
517 	    str = stackblock() + strloc;
518 
519 	switch (subtype) {
520 	case VSASSIGN:
521 		setvar(str, startp, 0);
522 		amount = startp - expdest;
523 		STADJUST(amount, expdest);
524 		varflags &= ~VSNUL;
525 		if (c != 0)
526 			*loc = c;
527 		return 1;
528 
529 	case VSQUESTION:
530 		if (*p != CTLENDVAR) {
531 			outfmt(&errout, "%s\n", startp);
532 			error((char *)NULL);
533 		}
534 		error("%.*s: parameter %snot set", p - str - 1,
535 		      str, (varflags & VSNUL) ? "null or "
536 					      : nullstr);
537 		return 0;
538 
539 	case VSTRIMLEFT:
540 		for (loc = startp; loc < str; loc++) {
541 			c = *loc;
542 			*loc = '\0';
543 			if (patmatch(str, startp, varflags & VSQUOTE)) {
544 				*loc = c;
545 				goto recordleft;
546 			}
547 			*loc = c;
548 			if ((varflags & VSQUOTE) && *loc == CTLESC)
549 				loc++;
550 		}
551 		return 0;
552 
553 	case VSTRIMLEFTMAX:
554 		for (loc = str - 1; loc >= startp;) {
555 			c = *loc;
556 			*loc = '\0';
557 			if (patmatch(str, startp, varflags & VSQUOTE)) {
558 				*loc = c;
559 				goto recordleft;
560 			}
561 			*loc = c;
562 			loc--;
563 			if ((varflags & VSQUOTE) && loc > startp &&
564 			    *(loc - 1) == CTLESC) {
565 				for (q = startp; q < loc; q++)
566 					if (*q == CTLESC)
567 						q++;
568 				if (q > loc)
569 					loc--;
570 			}
571 		}
572 		return 0;
573 
574 	case VSTRIMRIGHT:
575 		for (loc = str - 1; loc >= startp;) {
576 			if (patmatch(str, loc, varflags & VSQUOTE)) {
577 				amount = loc - expdest;
578 				STADJUST(amount, expdest);
579 				return 1;
580 			}
581 			loc--;
582 			if ((varflags & VSQUOTE) && loc > startp &&
583 			    *(loc - 1) == CTLESC) {
584 				for (q = startp; q < loc; q++)
585 					if (*q == CTLESC)
586 						q++;
587 				if (q > loc)
588 					loc--;
589 			}
590 		}
591 		return 0;
592 
593 	case VSTRIMRIGHTMAX:
594 		for (loc = startp; loc < str - 1; loc++) {
595 			if (patmatch(str, loc, varflags & VSQUOTE)) {
596 				amount = loc - expdest;
597 				STADJUST(amount, expdest);
598 				return 1;
599 			}
600 			if ((varflags & VSQUOTE) && *loc == CTLESC)
601 				loc++;
602 		}
603 		return 0;
604 
605 
606 	default:
607 		abort();
608 	}
609 
610 recordleft:
611 	amount = ((str - 1) - (loc - startp)) - expdest;
612 	STADJUST(amount, expdest);
613 	while (loc != str - 1)
614 		*startp++ = *loc++;
615 	return 1;
616 }
617 
618 
619 /*
620  * Expand a variable, and return a pointer to the next character in the
621  * input string.
622  */
623 
624 STATIC char *
625 evalvar(char *p, int flag)
626 {
627 	int subtype;
628 	int varflags;
629 	char *var;
630 	char *val;
631 	int patloc;
632 	int c;
633 	int set;
634 	int special;
635 	int startloc;
636 	int varlen;
637 	int easy;
638 	int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR);
639 
640 	varflags = *p++;
641 	subtype = varflags & VSTYPE;
642 	var = p;
643 	special = 0;
644 	if (! is_name(*p))
645 		special = 1;
646 	p = strchr(p, '=') + 1;
647 again: /* jump here after setting a variable with ${var=text} */
648 	if (special) {
649 		set = varisset(var, varflags & VSNUL);
650 		val = NULL;
651 	} else {
652 		val = bltinlookup(var, 1);
653 		if (val == NULL || ((varflags & VSNUL) && val[0] == '\0')) {
654 			val = NULL;
655 			set = 0;
656 		} else
657 			set = 1;
658 	}
659 	varlen = 0;
660 	startloc = expdest - stackblock();
661 	if (!set && uflag) {
662 		switch (subtype) {
663 		case VSNORMAL:
664 		case VSTRIMLEFT:
665 		case VSTRIMLEFTMAX:
666 		case VSTRIMRIGHT:
667 		case VSTRIMRIGHTMAX:
668 		case VSLENGTH:
669 			error("%.*s: parameter not set", p - var - 1, var);
670 		}
671 	}
672 	if (set && subtype != VSPLUS) {
673 		/* insert the value of the variable */
674 		if (special) {
675 			varvalue(var, varflags & VSQUOTE, flag & EXP_FULL);
676 			if (subtype == VSLENGTH) {
677 				varlen = expdest - stackblock() - startloc;
678 				STADJUST(-varlen, expdest);
679 			}
680 		} else {
681 			char const *syntax = (varflags & VSQUOTE) ? DQSYNTAX
682 								  : BASESYNTAX;
683 
684 			if (subtype == VSLENGTH) {
685 				for (;*val; val++)
686 					varlen++;
687 			}
688 			else {
689 				while (*val) {
690 					if (quotes &&
691 					    syntax[(int)*val] == CCTL)
692 						STPUTC(CTLESC, expdest);
693 					STPUTC(*val++, expdest);
694 				}
695 
696 			}
697 		}
698 	}
699 
700 	if (subtype == VSPLUS)
701 		set = ! set;
702 
703 	easy = ((varflags & VSQUOTE) == 0 ||
704 		(*var == '@' && shellparam.nparam != 1));
705 
706 
707 	switch (subtype) {
708 	case VSLENGTH:
709 		expdest = cvtnum(varlen, expdest);
710 		goto record;
711 
712 	case VSNORMAL:
713 		if (!easy)
714 			break;
715 record:
716 		recordregion(startloc, expdest - stackblock(),
717 			     varflags & VSQUOTE);
718 		break;
719 
720 	case VSPLUS:
721 	case VSMINUS:
722 		if (!set) {
723 			argstr(p, flag);
724 			break;
725 		}
726 		if (easy)
727 			goto record;
728 		break;
729 
730 	case VSTRIMLEFT:
731 	case VSTRIMLEFTMAX:
732 	case VSTRIMRIGHT:
733 	case VSTRIMRIGHTMAX:
734 		if (!set)
735 			break;
736 		/*
737 		 * Terminate the string and start recording the pattern
738 		 * right after it
739 		 */
740 		STPUTC('\0', expdest);
741 		patloc = expdest - stackblock();
742 		if (subevalvar(p, NULL, patloc, subtype,
743 			       startloc, varflags) == 0) {
744 			int amount = (expdest - stackblock() - patloc) + 1;
745 			STADJUST(-amount, expdest);
746 		}
747 		/* Remove any recorded regions beyond start of variable */
748 		removerecordregions(startloc);
749 		goto record;
750 
751 	case VSASSIGN:
752 	case VSQUESTION:
753 		if (!set) {
754 			if (subevalvar(p, var, 0, subtype, startloc, varflags)) {
755 				varflags &= ~VSNUL;
756 				/*
757 				 * Remove any recorded regions beyond
758 				 * start of variable
759 				 */
760 				removerecordregions(startloc);
761 				goto again;
762 			}
763 			break;
764 		}
765 		if (easy)
766 			goto record;
767 		break;
768 
769 	default:
770 		abort();
771 	}
772 
773 	if (subtype != VSNORMAL) {	/* skip to end of alternative */
774 		int nesting = 1;
775 		for (;;) {
776 			if ((c = *p++) == CTLESC)
777 				p++;
778 			else if (c == CTLBACKQ || c == (CTLBACKQ|CTLQUOTE)) {
779 				if (set)
780 					argbackq = argbackq->next;
781 			} else if (c == CTLVAR) {
782 				if ((*p++ & VSTYPE) != VSNORMAL)
783 					nesting++;
784 			} else if (c == CTLENDVAR) {
785 				if (--nesting == 0)
786 					break;
787 			}
788 		}
789 	}
790 	return p;
791 }
792 
793 
794 
795 /*
796  * Test whether a specialized variable is set.
797  */
798 
799 STATIC int
800 varisset(char *name, int nulok)
801 {
802 
803 	if (*name == '!')
804 		return backgndpid != -1;
805 	else if (*name == '@' || *name == '*') {
806 		if (*shellparam.p == NULL)
807 			return 0;
808 
809 		if (nulok) {
810 			char **av;
811 
812 			for (av = shellparam.p; *av; av++)
813 				if (**av != '\0')
814 					return 1;
815 			return 0;
816 		}
817 	} else if (is_digit(*name)) {
818 		char *ap;
819 		int num = atoi(name);
820 
821 		if (num > shellparam.nparam)
822 			return 0;
823 
824 		if (num == 0)
825 			ap = arg0;
826 		else
827 			ap = shellparam.p[num - 1];
828 
829 		if (nulok && (ap == NULL || *ap == '\0'))
830 			return 0;
831 	}
832 	return 1;
833 }
834 
835 
836 
837 /*
838  * Add the value of a specialized variable to the stack string.
839  */
840 
841 STATIC void
842 varvalue(char *name, int quoted, int allow_split)
843 {
844 	int num;
845 	char *p;
846 	int i;
847 	extern int oexitstatus;
848 	char sep;
849 	char **ap;
850 	char const *syntax;
851 
852 #define STRTODEST(p) \
853 	do {\
854 	if (allow_split) { \
855 		syntax = quoted? DQSYNTAX : BASESYNTAX; \
856 		while (*p) { \
857 			if (syntax[(int)*p] == CCTL) \
858 				STPUTC(CTLESC, expdest); \
859 			STPUTC(*p++, expdest); \
860 		} \
861 	} else \
862 		while (*p) \
863 			STPUTC(*p++, expdest); \
864 	} while (0)
865 
866 
867 	switch (*name) {
868 	case '$':
869 		num = rootpid;
870 		goto numvar;
871 	case '?':
872 		num = oexitstatus;
873 		goto numvar;
874 	case '#':
875 		num = shellparam.nparam;
876 		goto numvar;
877 	case '!':
878 		num = backgndpid;
879 numvar:
880 		expdest = cvtnum(num, expdest);
881 		break;
882 	case '-':
883 		for (i = 0 ; i < NOPTS ; i++) {
884 			if (optlist[i].val)
885 				STPUTC(optlist[i].letter, expdest);
886 		}
887 		break;
888 	case '@':
889 		if (allow_split && quoted) {
890 			for (ap = shellparam.p ; (p = *ap++) != NULL ; ) {
891 				STRTODEST(p);
892 				if (*ap)
893 					STPUTC('\0', expdest);
894 			}
895 			break;
896 		}
897 		/* FALLTHROUGH */
898 	case '*':
899 		if (ifsset() != 0)
900 			sep = ifsval()[0];
901 		else
902 			sep = ' ';
903 		for (ap = shellparam.p ; (p = *ap++) != NULL ; ) {
904 			STRTODEST(p);
905 			if (*ap && sep)
906 				STPUTC(sep, expdest);
907 		}
908 		break;
909 	case '0':
910 		p = arg0;
911 		STRTODEST(p);
912 		break;
913 	default:
914 		if (is_digit(*name)) {
915 			num = atoi(name);
916 			if (num > 0 && num <= shellparam.nparam) {
917 				p = shellparam.p[num - 1];
918 				STRTODEST(p);
919 			}
920 		}
921 		break;
922 	}
923 }
924 
925 
926 
927 /*
928  * Record the the fact that we have to scan this region of the
929  * string for IFS characters.
930  */
931 
932 STATIC void
933 recordregion(int start, int end, int nulonly)
934 {
935 	struct ifsregion *ifsp;
936 
937 	if (ifslastp == NULL) {
938 		ifsp = &ifsfirst;
939 	} else {
940 		ifsp = (struct ifsregion *)ckmalloc(sizeof (struct ifsregion));
941 		ifslastp->next = ifsp;
942 	}
943 	ifslastp = ifsp;
944 	ifslastp->next = NULL;
945 	ifslastp->begoff = start;
946 	ifslastp->endoff = end;
947 	ifslastp->nulonly = nulonly;
948 }
949 
950 
951 
952 /*
953  * Break the argument string into pieces based upon IFS and add the
954  * strings to the argument list.  The regions of the string to be
955  * searched for IFS characters have been stored by recordregion.
956  */
957 STATIC void
958 ifsbreakup(char *string, struct arglist *arglist)
959 {
960 	struct ifsregion *ifsp;
961 	struct strlist *sp;
962 	char *start;
963 	char *p;
964 	char *q;
965 	char *ifs;
966 	int ifsspc;
967 	int nulonly;
968 
969 
970 	start = string;
971 	ifsspc = 0;
972 	nulonly = 0;
973 	if (ifslastp != NULL) {
974 		ifsp = &ifsfirst;
975 		do {
976 			p = string + ifsp->begoff;
977 			nulonly = ifsp->nulonly;
978 			ifs = nulonly ? nullstr :
979 				( ifsset() ? ifsval() : " \t\n" );
980 			ifsspc = 0;
981 			while (p < string + ifsp->endoff) {
982 				q = p;
983 				if (*p == CTLESC)
984 					p++;
985 				if (strchr(ifs, *p)) {
986 					if (!nulonly)
987 						ifsspc = (strchr(" \t\n", *p) != NULL);
988 					/* Ignore IFS whitespace at start */
989 					if (q == start && ifsspc) {
990 						p++;
991 						start = p;
992 						continue;
993 					}
994 					*q = '\0';
995 					sp = (struct strlist *)stalloc(sizeof *sp);
996 					sp->text = start;
997 					*arglist->lastp = sp;
998 					arglist->lastp = &sp->next;
999 					p++;
1000 					if (!nulonly) {
1001 						for (;;) {
1002 							if (p >= string + ifsp->endoff) {
1003 								break;
1004 							}
1005 							q = p;
1006 							if (*p == CTLESC)
1007 								p++;
1008 							if (strchr(ifs, *p) == NULL ) {
1009 								p = q;
1010 								break;
1011 							} else if (strchr(" \t\n",*p) == NULL) {
1012 								if (ifsspc) {
1013 									p++;
1014 									ifsspc = 0;
1015 								} else {
1016 									p = q;
1017 									break;
1018 								}
1019 							} else
1020 								p++;
1021 						}
1022 					}
1023 					start = p;
1024 				} else
1025 					p++;
1026 			}
1027 		} while ((ifsp = ifsp->next) != NULL);
1028 		if (*start || (!ifsspc && start > string &&
1029 			(nulonly || 1))) {
1030 			sp = (struct strlist *)stalloc(sizeof *sp);
1031 			sp->text = start;
1032 			*arglist->lastp = sp;
1033 			arglist->lastp = &sp->next;
1034 		}
1035 	} else {
1036 		sp = (struct strlist *)stalloc(sizeof *sp);
1037 		sp->text = start;
1038 		*arglist->lastp = sp;
1039 		arglist->lastp = &sp->next;
1040 	}
1041 }
1042 
1043 
1044 
1045 /*
1046  * Expand shell metacharacters.  At this point, the only control characters
1047  * should be escapes.  The results are stored in the list exparg.
1048  */
1049 
1050 char *expdir;
1051 
1052 
1053 STATIC void
1054 expandmeta(struct strlist *str, int flag __unused)
1055 {
1056 	char *p;
1057 	struct strlist **savelastp;
1058 	struct strlist *sp;
1059 	char c;
1060 	/* TODO - EXP_REDIR */
1061 
1062 	while (str) {
1063 		if (fflag)
1064 			goto nometa;
1065 		p = str->text;
1066 		for (;;) {			/* fast check for meta chars */
1067 			if ((c = *p++) == '\0')
1068 				goto nometa;
1069 			if (c == '*' || c == '?' || c == '[' || c == '!')
1070 				break;
1071 		}
1072 		savelastp = exparg.lastp;
1073 		INTOFF;
1074 		if (expdir == NULL) {
1075 			int i = strlen(str->text);
1076 			expdir = ckmalloc(i < 2048 ? 2048 : i); /* XXX */
1077 		}
1078 
1079 		expmeta(expdir, str->text);
1080 		ckfree(expdir);
1081 		expdir = NULL;
1082 		INTON;
1083 		if (exparg.lastp == savelastp) {
1084 			/*
1085 			 * no matches
1086 			 */
1087 nometa:
1088 			*exparg.lastp = str;
1089 			rmescapes(str->text);
1090 			exparg.lastp = &str->next;
1091 		} else {
1092 			*exparg.lastp = NULL;
1093 			*savelastp = sp = expsort(*savelastp);
1094 			while (sp->next != NULL)
1095 				sp = sp->next;
1096 			exparg.lastp = &sp->next;
1097 		}
1098 		str = str->next;
1099 	}
1100 }
1101 
1102 
1103 /*
1104  * Do metacharacter (i.e. *, ?, [...]) expansion.
1105  */
1106 
1107 STATIC void
1108 expmeta(char *enddir, char *name)
1109 {
1110 	char *p;
1111 	char *q;
1112 	char *start;
1113 	char *endname;
1114 	int metaflag;
1115 	struct stat statb;
1116 	DIR *dirp;
1117 	struct dirent *dp;
1118 	int atend;
1119 	int matchdot;
1120 
1121 	metaflag = 0;
1122 	start = name;
1123 	for (p = name ; ; p++) {
1124 		if (*p == '*' || *p == '?')
1125 			metaflag = 1;
1126 		else if (*p == '[') {
1127 			q = p + 1;
1128 			if (*q == '!' || *q == '^')
1129 				q++;
1130 			for (;;) {
1131 				while (*q == CTLQUOTEMARK)
1132 					q++;
1133 				if (*q == CTLESC)
1134 					q++;
1135 				if (*q == '/' || *q == '\0')
1136 					break;
1137 				if (*++q == ']') {
1138 					metaflag = 1;
1139 					break;
1140 				}
1141 			}
1142 		} else if (*p == '!' && p[1] == '!'	&& (p == name || p[-1] == '/')) {
1143 			metaflag = 1;
1144 		} else if (*p == '\0')
1145 			break;
1146 		else if (*p == CTLQUOTEMARK)
1147 			continue;
1148 		else if (*p == CTLESC)
1149 			p++;
1150 		if (*p == '/') {
1151 			if (metaflag)
1152 				break;
1153 			start = p + 1;
1154 		}
1155 	}
1156 	if (metaflag == 0) {	/* we've reached the end of the file name */
1157 		if (enddir != expdir)
1158 			metaflag++;
1159 		for (p = name ; ; p++) {
1160 			if (*p == CTLQUOTEMARK)
1161 				continue;
1162 			if (*p == CTLESC)
1163 				p++;
1164 			*enddir++ = *p;
1165 			if (*p == '\0')
1166 				break;
1167 		}
1168 		if (metaflag == 0 || stat(expdir, &statb) >= 0)
1169 			addfname(expdir);
1170 		return;
1171 	}
1172 	endname = p;
1173 	if (start != name) {
1174 		p = name;
1175 		while (p < start) {
1176 			while (*p == CTLQUOTEMARK)
1177 				p++;
1178 			if (*p == CTLESC)
1179 				p++;
1180 			*enddir++ = *p++;
1181 		}
1182 	}
1183 	if (enddir == expdir) {
1184 		p = ".";
1185 	} else if (enddir == expdir + 1 && *expdir == '/') {
1186 		p = "/";
1187 	} else {
1188 		p = expdir;
1189 		enddir[-1] = '\0';
1190 	}
1191 	if ((dirp = opendir(p)) == NULL)
1192 		return;
1193 	if (enddir != expdir)
1194 		enddir[-1] = '/';
1195 	if (*endname == 0) {
1196 		atend = 1;
1197 	} else {
1198 		atend = 0;
1199 		*endname++ = '\0';
1200 	}
1201 	matchdot = 0;
1202 	p = start;
1203 	while (*p == CTLQUOTEMARK)
1204 		p++;
1205 	if (*p == CTLESC)
1206 		p++;
1207 	if (*p == '.')
1208 		matchdot++;
1209 	while (! int_pending() && (dp = readdir(dirp)) != NULL) {
1210 		if (dp->d_name[0] == '.' && ! matchdot)
1211 			continue;
1212 		if (patmatch(start, dp->d_name, 0)) {
1213 			if (atend) {
1214 				scopy(dp->d_name, enddir);
1215 				addfname(expdir);
1216 			} else {
1217 				char *q;
1218 				for (p = enddir, q = dp->d_name;
1219 				     (*p++ = *q++) != '\0';)
1220 					continue;
1221 				p[-1] = '/';
1222 				expmeta(p, endname);
1223 			}
1224 		}
1225 	}
1226 	closedir(dirp);
1227 	if (! atend)
1228 		endname[-1] = '/';
1229 }
1230 
1231 
1232 /*
1233  * Add a file name to the list.
1234  */
1235 
1236 STATIC void
1237 addfname(char *name)
1238 {
1239 	char *p;
1240 	struct strlist *sp;
1241 
1242 	p = stalloc(strlen(name) + 1);
1243 	scopy(name, p);
1244 	sp = (struct strlist *)stalloc(sizeof *sp);
1245 	sp->text = p;
1246 	*exparg.lastp = sp;
1247 	exparg.lastp = &sp->next;
1248 }
1249 
1250 
1251 /*
1252  * Sort the results of file name expansion.  It calculates the number of
1253  * strings to sort and then calls msort (short for merge sort) to do the
1254  * work.
1255  */
1256 
1257 STATIC struct strlist *
1258 expsort(struct strlist *str)
1259 {
1260 	int len;
1261 	struct strlist *sp;
1262 
1263 	len = 0;
1264 	for (sp = str ; sp ; sp = sp->next)
1265 		len++;
1266 	return msort(str, len);
1267 }
1268 
1269 
1270 STATIC struct strlist *
1271 msort(struct strlist *list, int len)
1272 {
1273 	struct strlist *p, *q = NULL;
1274 	struct strlist **lpp;
1275 	int half;
1276 	int n;
1277 
1278 	if (len <= 1)
1279 		return list;
1280 	half = len >> 1;
1281 	p = list;
1282 	for (n = half ; --n >= 0 ; ) {
1283 		q = p;
1284 		p = p->next;
1285 	}
1286 	q->next = NULL;			/* terminate first half of list */
1287 	q = msort(list, half);		/* sort first half of list */
1288 	p = msort(p, len - half);		/* sort second half */
1289 	lpp = &list;
1290 	for (;;) {
1291 		if (strcmp(p->text, q->text) < 0) {
1292 			*lpp = p;
1293 			lpp = &p->next;
1294 			if ((p = *lpp) == NULL) {
1295 				*lpp = q;
1296 				break;
1297 			}
1298 		} else {
1299 			*lpp = q;
1300 			lpp = &q->next;
1301 			if ((q = *lpp) == NULL) {
1302 				*lpp = p;
1303 				break;
1304 			}
1305 		}
1306 	}
1307 	return list;
1308 }
1309 
1310 
1311 
1312 /*
1313  * Returns true if the pattern matches the string.
1314  */
1315 
1316 int
1317 patmatch(char *pattern, char *string, int squoted)
1318 {
1319 #ifdef notdef
1320 	if (pattern[0] == '!' && pattern[1] == '!')
1321 		return 1 - pmatch(pattern + 2, string);
1322 	else
1323 #endif
1324 		return pmatch(pattern, string, squoted);
1325 }
1326 
1327 
1328 STATIC int
1329 pmatch(char *pattern, char *string, int squoted)
1330 {
1331 	char *p, *q;
1332 	char c;
1333 
1334 	p = pattern;
1335 	q = string;
1336 	for (;;) {
1337 		switch (c = *p++) {
1338 		case '\0':
1339 			goto breakloop;
1340 		case CTLESC:
1341 			if (squoted && *q == CTLESC)
1342 				q++;
1343 			if (*q++ != *p++)
1344 				return 0;
1345 			break;
1346 		case CTLQUOTEMARK:
1347 			continue;
1348 		case '?':
1349 			if (squoted && *q == CTLESC)
1350 				q++;
1351 			if (*q++ == '\0')
1352 				return 0;
1353 			break;
1354 		case '*':
1355 			c = *p;
1356 			while (c == CTLQUOTEMARK || c == '*')
1357 				c = *++p;
1358 			if (c != CTLESC &&  c != CTLQUOTEMARK &&
1359 			    c != '?' && c != '*' && c != '[') {
1360 				while (*q != c) {
1361 					if (squoted && *q == CTLESC &&
1362 					    q[1] == c)
1363 						break;
1364 					if (*q == '\0')
1365 						return 0;
1366 					if (squoted && *q == CTLESC)
1367 						q++;
1368 					q++;
1369 				}
1370 			}
1371 			do {
1372 				if (pmatch(p, q, squoted))
1373 					return 1;
1374 				if (squoted && *q == CTLESC)
1375 					q++;
1376 			} while (*q++ != '\0');
1377 			return 0;
1378 		case '[': {
1379 			char *endp;
1380 			int invert, found;
1381 			char chr;
1382 
1383 			endp = p;
1384 			if (*endp == '!' || *endp == '^')
1385 				endp++;
1386 			for (;;) {
1387 				while (*endp == CTLQUOTEMARK)
1388 					endp++;
1389 				if (*endp == '\0')
1390 					goto dft;		/* no matching ] */
1391 				if (*endp == CTLESC)
1392 					endp++;
1393 				if (*++endp == ']')
1394 					break;
1395 			}
1396 			invert = 0;
1397 			if (*p == '!' || *p == '^') {
1398 				invert++;
1399 				p++;
1400 			}
1401 			found = 0;
1402 			chr = *q++;
1403 			if (squoted && chr == CTLESC)
1404 				chr = *q++;
1405 			if (chr == '\0')
1406 				return 0;
1407 			c = *p++;
1408 			do {
1409 				if (c == CTLQUOTEMARK)
1410 					continue;
1411 				if (c == CTLESC)
1412 					c = *p++;
1413 				if (*p == '-' && p[1] != ']') {
1414 					p++;
1415 					while (*p == CTLQUOTEMARK)
1416 						p++;
1417 					if (*p == CTLESC)
1418 						p++;
1419 					if (   collate_range_cmp(chr, c) >= 0
1420 					    && collate_range_cmp(chr, *p) <= 0
1421 					   )
1422 						found = 1;
1423 					p++;
1424 				} else {
1425 					if (chr == c)
1426 						found = 1;
1427 				}
1428 			} while ((c = *p++) != ']');
1429 			if (found == invert)
1430 				return 0;
1431 			break;
1432 		}
1433 dft:	        default:
1434 			if (squoted && *q == CTLESC)
1435 				q++;
1436 			if (*q++ != c)
1437 				return 0;
1438 			break;
1439 		}
1440 	}
1441 breakloop:
1442 	if (*q != '\0')
1443 		return 0;
1444 	return 1;
1445 }
1446 
1447 
1448 
1449 /*
1450  * Remove any CTLESC characters from a string.
1451  */
1452 
1453 void
1454 rmescapes(char *str)
1455 {
1456 	char *p, *q;
1457 
1458 	p = str;
1459 	while (*p != CTLESC && *p != CTLQUOTEMARK) {
1460 		if (*p++ == '\0')
1461 			return;
1462 	}
1463 	q = p;
1464 	while (*p) {
1465 		if (*p == CTLQUOTEMARK) {
1466 			p++;
1467 			continue;
1468 		}
1469 		if (*p == CTLESC)
1470 			p++;
1471 		*q++ = *p++;
1472 	}
1473 	*q = '\0';
1474 }
1475 
1476 
1477 
1478 /*
1479  * See if a pattern matches in a case statement.
1480  */
1481 
1482 int
1483 casematch(union node *pattern, char *val)
1484 {
1485 	struct stackmark smark;
1486 	int result;
1487 	char *p;
1488 
1489 	setstackmark(&smark);
1490 	argbackq = pattern->narg.backquote;
1491 	STARTSTACKSTR(expdest);
1492 	ifslastp = NULL;
1493 	argstr(pattern->narg.text, EXP_TILDE | EXP_CASE);
1494 	STPUTC('\0', expdest);
1495 	p = grabstackstr(expdest);
1496 	result = patmatch(p, val, 0);
1497 	popstackmark(&smark);
1498 	return result;
1499 }
1500 
1501 /*
1502  * Our own itoa().
1503  */
1504 
1505 STATIC char *
1506 cvtnum(int num, char *buf)
1507 {
1508 	char temp[32];
1509 	int neg = num < 0;
1510 	char *p = temp + 31;
1511 
1512 	temp[31] = '\0';
1513 
1514 	do {
1515 		*--p = num % 10 + '0';
1516 	} while ((num /= 10) != 0);
1517 
1518 	if (neg)
1519 		*--p = '-';
1520 
1521 	while (*p)
1522 		STPUTC(*p++, buf);
1523 	return buf;
1524 }
1525 
1526 /*
1527  * Do most of the work for wordexp(3).
1528  */
1529 
1530 int
1531 wordexpcmd(int argc, char **argv)
1532 {
1533 	size_t len;
1534 	int i;
1535 
1536 	out1fmt("%08x", argc - 1);
1537 	for (i = 1, len = 0; i < argc; i++)
1538 		len += strlen(argv[i]);
1539 	out1fmt("%08x", (int)len);
1540 	for (i = 1; i < argc; i++) {
1541 		out1str(argv[i]);
1542 		out1c('\0');
1543 	}
1544         return (0);
1545 }
1546