xref: /freebsd/bin/sh/expand.c (revision 68e7a217f8019b955f87547f218e95ab237597af)
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 static const char rcsid[] =
42   "$FreeBSD$";
43 #endif /* not lint */
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 
56 /*
57  * Routines to expand arguments to commands.  We have to deal with
58  * backquotes, shell variables, and file metacharacters.
59  */
60 
61 #include "shell.h"
62 #include "main.h"
63 #include "nodes.h"
64 #include "eval.h"
65 #include "expand.h"
66 #include "syntax.h"
67 #include "parser.h"
68 #include "jobs.h"
69 #include "options.h"
70 #include "var.h"
71 #include "input.h"
72 #include "output.h"
73 #include "memalloc.h"
74 #include "error.h"
75 #include "mystring.h"
76 #include "arith.h"
77 #include "show.h"
78 
79 /*
80  * Structure specifying which parts of the string should be searched
81  * for IFS characters.
82  */
83 
84 struct ifsregion {
85 	struct ifsregion *next;	/* next region in list */
86 	int begoff;		/* offset of start of region */
87 	int endoff;		/* offset of end of region */
88 	int nulonly;		/* search for nul bytes only */
89 };
90 
91 
92 char *expdest;			/* output of current string */
93 struct nodelist *argbackq;	/* list of back quote expressions */
94 struct ifsregion ifsfirst;	/* first struct in list of ifs regions */
95 struct ifsregion *ifslastp;	/* last struct in list */
96 struct arglist exparg;		/* holds expanded arg list */
97 
98 STATIC void argstr(char *, int);
99 STATIC char *exptilde(char *, int);
100 STATIC void expbackq(union node *, int, int);
101 STATIC int subevalvar(char *, char *, int, int, int, int);
102 STATIC char *evalvar(char *, int);
103 STATIC int varisset(char *, int);
104 STATIC void varvalue(char *, int, int);
105 STATIC void recordregion(int, int, int);
106 STATIC void removerecordregions(int);
107 STATIC void ifsbreakup(char *, struct arglist *);
108 STATIC void expandmeta(struct strlist *, int);
109 STATIC void expmeta(char *, char *);
110 STATIC void addfname(char *);
111 STATIC struct strlist *expsort(struct strlist *);
112 STATIC struct strlist *msort(struct strlist *, int);
113 STATIC int pmatch(char *, char *, int);
114 STATIC char *cvtnum(int, char *);
115 STATIC int collate_range_cmp(int, int);
116 
117 STATIC int
118 collate_range_cmp (int c1, int c2)
119 {
120 	static char s1[2], s2[2];
121 	int ret;
122 
123 	c1 &= UCHAR_MAX;
124 	c2 &= UCHAR_MAX;
125 	if (c1 == c2)
126 		return (0);
127 	s1[0] = c1;
128 	s2[0] = c2;
129 	if ((ret = strcoll(s1, s2)) != 0)
130 		return (ret);
131 	return (c1 - c2);
132 }
133 
134 /*
135  * Expand shell variables and backquotes inside a here document.
136  *	union node *arg		the document
137  *	int fd;			where to write the expanded version
138  */
139 
140 void
141 expandhere(union node *arg, int fd)
142 {
143 	herefd = fd;
144 	expandarg(arg, (struct arglist *)NULL, 0);
145 	xwrite(fd, stackblock(), expdest - stackblock());
146 }
147 
148 
149 /*
150  * Perform variable substitution and command substitution on an argument,
151  * placing the resulting list of arguments in arglist.  If EXP_FULL is true,
152  * perform splitting and file name expansion.  When arglist is NULL, perform
153  * here document expansion.
154  */
155 
156 void
157 expandarg(union node *arg, struct arglist *arglist, int flag)
158 {
159 	struct strlist *sp;
160 	char *p;
161 
162 	argbackq = arg->narg.backquote;
163 	STARTSTACKSTR(expdest);
164 	ifsfirst.next = NULL;
165 	ifslastp = NULL;
166 	argstr(arg->narg.text, flag);
167 	if (arglist == NULL) {
168 		return;			/* here document expanded */
169 	}
170 	STPUTC('\0', expdest);
171 	p = grabstackstr(expdest);
172 	exparg.lastp = &exparg.list;
173 	/*
174 	 * TODO - EXP_REDIR
175 	 */
176 	if (flag & EXP_FULL) {
177 		ifsbreakup(p, &exparg);
178 		*exparg.lastp = NULL;
179 		exparg.lastp = &exparg.list;
180 		expandmeta(exparg.list, flag);
181 	} else {
182 		if (flag & EXP_REDIR) /*XXX - for now, just remove escapes */
183 			rmescapes(p);
184 		sp = (struct strlist *)stalloc(sizeof (struct strlist));
185 		sp->text = p;
186 		*exparg.lastp = sp;
187 		exparg.lastp = &sp->next;
188 	}
189 	while (ifsfirst.next != NULL) {
190 		struct ifsregion *ifsp;
191 		INTOFF;
192 		ifsp = ifsfirst.next->next;
193 		ckfree(ifsfirst.next);
194 		ifsfirst.next = ifsp;
195 		INTON;
196 	}
197 	*exparg.lastp = NULL;
198 	if (exparg.list) {
199 		*arglist->lastp = exparg.list;
200 		arglist->lastp = exparg.lastp;
201 	}
202 }
203 
204 
205 
206 /*
207  * Perform variable and command substitution.  If EXP_FULL is set, output CTLESC
208  * characters to allow for further processing.  Otherwise treat
209  * $@ like $* since no splitting will be performed.
210  */
211 
212 STATIC void
213 argstr(char *p, int flag)
214 {
215 	char c;
216 	int quotes = flag & (EXP_FULL | EXP_CASE);	/* do CTLESC */
217 	int firsteq = 1;
218 
219 	if (*p == '~' && (flag & (EXP_TILDE | EXP_VARTILDE)))
220 		p = exptilde(p, flag);
221 	for (;;) {
222 		switch (c = *p++) {
223 		case '\0':
224 		case CTLENDVAR: /* ??? */
225 			goto breakloop;
226 		case CTLQUOTEMARK:
227 			/* "$@" syntax adherence hack */
228 			if (p[0] == CTLVAR && p[2] == '@' && p[3] == '=')
229 				break;
230 			if ((flag & EXP_FULL) != 0)
231 				STPUTC(c, expdest);
232 			break;
233 		case CTLESC:
234 			if (quotes)
235 				STPUTC(c, expdest);
236 			c = *p++;
237 			STPUTC(c, expdest);
238 			break;
239 		case CTLVAR:
240 			p = evalvar(p, flag);
241 			break;
242 		case CTLBACKQ:
243 		case CTLBACKQ|CTLQUOTE:
244 			expbackq(argbackq->n, c & CTLQUOTE, flag);
245 			argbackq = argbackq->next;
246 			break;
247 		case CTLENDARI:
248 			expari(flag);
249 			break;
250 		case ':':
251 		case '=':
252 			/*
253 			 * sort of a hack - expand tildes in variable
254 			 * assignments (after the first '=' and after ':'s).
255 			 */
256 			STPUTC(c, expdest);
257 			if (flag & EXP_VARTILDE && *p == '~') {
258 				if (c == '=') {
259 					if (firsteq)
260 						firsteq = 0;
261 					else
262 						break;
263 				}
264 				p = exptilde(p, flag);
265 			}
266 			break;
267 		default:
268 			STPUTC(c, expdest);
269 		}
270 	}
271 breakloop:;
272 }
273 
274 STATIC char *
275 exptilde(char *p, int flag)
276 {
277 	char c, *startp = p;
278 	struct passwd *pw;
279 	char *home;
280 	int quotes = flag & (EXP_FULL | EXP_CASE);
281 
282 	while ((c = *p) != '\0') {
283 		switch(c) {
284 		case CTLESC:
285 			return (startp);
286 		case CTLQUOTEMARK:
287 			return (startp);
288 		case ':':
289 			if (flag & EXP_VARTILDE)
290 				goto done;
291 			break;
292 		case '/':
293 			goto done;
294 		}
295 		p++;
296 	}
297 done:
298 	*p = '\0';
299 	if (*(startp+1) == '\0') {
300 		if ((home = lookupvar("HOME")) == NULL)
301 			goto lose;
302 	} else {
303 		if ((pw = getpwnam(startp+1)) == NULL)
304 			goto lose;
305 		home = pw->pw_dir;
306 	}
307 	if (*home == '\0')
308 		goto lose;
309 	*p = c;
310 	while ((c = *home++) != '\0') {
311 		if (quotes && SQSYNTAX[(int)c] == CCTL)
312 			STPUTC(CTLESC, expdest);
313 		STPUTC(c, expdest);
314 	}
315 	return (p);
316 lose:
317 	*p = c;
318 	return (startp);
319 }
320 
321 
322 STATIC void
323 removerecordregions(int endoff)
324 {
325 	if (ifslastp == NULL)
326 		return;
327 
328 	if (ifsfirst.endoff > endoff) {
329 		while (ifsfirst.next != NULL) {
330 			struct ifsregion *ifsp;
331 			INTOFF;
332 			ifsp = ifsfirst.next->next;
333 			ckfree(ifsfirst.next);
334 			ifsfirst.next = ifsp;
335 			INTON;
336 		}
337 		if (ifsfirst.begoff > endoff)
338 			ifslastp = NULL;
339 		else {
340 			ifslastp = &ifsfirst;
341 			ifsfirst.endoff = endoff;
342 		}
343 		return;
344 	}
345 
346 	ifslastp = &ifsfirst;
347 	while (ifslastp->next && ifslastp->next->begoff < endoff)
348 		ifslastp=ifslastp->next;
349 	while (ifslastp->next != NULL) {
350 		struct ifsregion *ifsp;
351 		INTOFF;
352 		ifsp = ifslastp->next->next;
353 		ckfree(ifslastp->next);
354 		ifslastp->next = ifsp;
355 		INTON;
356 	}
357 	if (ifslastp->endoff > endoff)
358 		ifslastp->endoff = endoff;
359 }
360 
361 /*
362  * Expand arithmetic expression.  Backup to start of expression,
363  * evaluate, place result in (backed up) result, adjust string position.
364  */
365 void
366 expari(int flag)
367 {
368 	char *p, *start;
369 	int result;
370 	int begoff;
371 	int quotes = flag & (EXP_FULL | EXP_CASE);
372 	int quoted;
373 
374 
375 	/*
376 	 * This routine is slightly over-complicated for
377 	 * efficiency.  First we make sure there is
378 	 * enough space for the result, which may be bigger
379 	 * than the expression if we add exponentiation.  Next we
380 	 * scan backwards looking for the start of arithmetic.  If the
381 	 * next previous character is a CTLESC character, then we
382 	 * have to rescan starting from the beginning since CTLESC
383 	 * characters have to be processed left to right.
384 	 */
385 #if INT_MAX / 1000000000 >= 10 || INT_MIN / 1000000000 <= -10
386 #error "integers with more than 10 digits are not supported"
387 #endif
388 	CHECKSTRSPACE(12 - 2, expdest);
389 	USTPUTC('\0', expdest);
390 	start = stackblock();
391 	p = expdest - 2;
392 	while (p >= start && *p != CTLARI)
393 		--p;
394 	if (p < start || *p != CTLARI)
395 		error("missing CTLARI (shouldn't happen)");
396 	if (p > start && *(p - 1) == CTLESC)
397 		for (p = start; *p != CTLARI; p++)
398 			if (*p == CTLESC)
399 				p++;
400 
401 	if (p[1] == '"')
402 		quoted=1;
403 	else
404 		quoted=0;
405 	begoff = p - start;
406 	removerecordregions(begoff);
407 	if (quotes)
408 		rmescapes(p+2);
409 	result = arith(p+2);
410 	fmtstr(p, 12, "%d", result);
411 	while (*p++)
412 		;
413 	if (quoted == 0)
414 		recordregion(begoff, p - 1 - start, 0);
415 	result = expdest - p + 1;
416 	STADJUST(-result, expdest);
417 }
418 
419 
420 /*
421  * Expand stuff in backwards quotes.
422  */
423 
424 STATIC void
425 expbackq(union node *cmd, int quoted, int flag)
426 {
427 	struct backcmd in;
428 	int i;
429 	char buf[128];
430 	char *p;
431 	char *dest = expdest;
432 	struct ifsregion saveifs, *savelastp;
433 	struct nodelist *saveargbackq;
434 	char lastc;
435 	int startloc = dest - stackblock();
436 	char const *syntax = quoted? DQSYNTAX : BASESYNTAX;
437 	int saveherefd;
438 	int quotes = flag & (EXP_FULL | EXP_CASE);
439 
440 	INTOFF;
441 	saveifs = ifsfirst;
442 	savelastp = ifslastp;
443 	saveargbackq = argbackq;
444 	saveherefd = herefd;
445 	herefd = -1;
446 	p = grabstackstr(dest);
447 	evalbackcmd(cmd, &in);
448 	ungrabstackstr(p, dest);
449 	ifsfirst = saveifs;
450 	ifslastp = savelastp;
451 	argbackq = saveargbackq;
452 	herefd = saveherefd;
453 
454 	p = in.buf;
455 	lastc = '\0';
456 	for (;;) {
457 		if (--in.nleft < 0) {
458 			if (in.fd < 0)
459 				break;
460 			while ((i = read(in.fd, buf, sizeof buf)) < 0 && errno == EINTR);
461 			TRACE(("expbackq: read returns %d\n", i));
462 			if (i <= 0)
463 				break;
464 			p = buf;
465 			in.nleft = i - 1;
466 		}
467 		lastc = *p++;
468 		if (lastc != '\0') {
469 			if (quotes && syntax[(int)lastc] == CCTL)
470 				STPUTC(CTLESC, dest);
471 			STPUTC(lastc, dest);
472 		}
473 	}
474 
475 	/* Eat all trailing newlines */
476 	for (p--; lastc == '\n'; lastc = *--p)
477 		STUNPUTC(dest);
478 
479 	if (in.fd >= 0)
480 		close(in.fd);
481 	if (in.buf)
482 		ckfree(in.buf);
483 	if (in.jp)
484 		exitstatus = waitforjob(in.jp, (int *)NULL);
485 	if (quoted == 0)
486 		recordregion(startloc, dest - stackblock(), 0);
487 	TRACE(("evalbackq: size=%d: \"%.*s\"\n",
488 		(dest - stackblock()) - startloc,
489 		(dest - stackblock()) - startloc,
490 		stackblock() + startloc));
491 	expdest = dest;
492 	INTON;
493 }
494 
495 
496 
497 STATIC int
498 subevalvar(char *p, char *str, int strloc, int subtype, int startloc,
499   int varflags)
500 {
501 	char *startp;
502 	char *loc = NULL;
503 	char *q;
504 	int c = 0;
505 	int saveherefd = herefd;
506 	struct nodelist *saveargbackq = argbackq;
507 	int amount;
508 
509 	herefd = -1;
510 	argstr(p, 0);
511 	STACKSTRNUL(expdest);
512 	herefd = saveherefd;
513 	argbackq = saveargbackq;
514 	startp = stackblock() + startloc;
515 	if (str == NULL)
516 	    str = stackblock() + strloc;
517 
518 	switch (subtype) {
519 	case VSASSIGN:
520 		setvar(str, startp, 0);
521 		amount = startp - expdest;
522 		STADJUST(amount, expdest);
523 		varflags &= ~VSNUL;
524 		if (c != 0)
525 			*loc = c;
526 		return 1;
527 
528 	case VSQUESTION:
529 		if (*p != CTLENDVAR) {
530 			outfmt(&errout, "%s\n", startp);
531 			error((char *)NULL);
532 		}
533 		error("%.*s: parameter %snot set", p - str - 1,
534 		      str, (varflags & VSNUL) ? "null or "
535 					      : nullstr);
536 		return 0;
537 
538 	case VSTRIMLEFT:
539 		for (loc = startp; loc < str; loc++) {
540 			c = *loc;
541 			*loc = '\0';
542 			if (patmatch(str, startp, varflags & VSQUOTE)) {
543 				*loc = c;
544 				goto recordleft;
545 			}
546 			*loc = c;
547 			if ((varflags & VSQUOTE) && *loc == CTLESC)
548 				loc++;
549 		}
550 		return 0;
551 
552 	case VSTRIMLEFTMAX:
553 		for (loc = str - 1; loc >= startp;) {
554 			c = *loc;
555 			*loc = '\0';
556 			if (patmatch(str, startp, varflags & VSQUOTE)) {
557 				*loc = c;
558 				goto recordleft;
559 			}
560 			*loc = c;
561 			loc--;
562 			if ((varflags & VSQUOTE) && loc > startp &&
563 			    *(loc - 1) == CTLESC) {
564 				for (q = startp; q < loc; q++)
565 					if (*q == CTLESC)
566 						q++;
567 				if (q > loc)
568 					loc--;
569 			}
570 		}
571 		return 0;
572 
573 	case VSTRIMRIGHT:
574 		for (loc = str - 1; loc >= startp;) {
575 			if (patmatch(str, loc, varflags & VSQUOTE)) {
576 				amount = loc - expdest;
577 				STADJUST(amount, expdest);
578 				return 1;
579 			}
580 			loc--;
581 			if ((varflags & VSQUOTE) && loc > startp &&
582 			    *(loc - 1) == CTLESC) {
583 				for (q = startp; q < loc; q++)
584 					if (*q == CTLESC)
585 						q++;
586 				if (q > loc)
587 					loc--;
588 			}
589 		}
590 		return 0;
591 
592 	case VSTRIMRIGHTMAX:
593 		for (loc = startp; loc < str - 1; loc++) {
594 			if (patmatch(str, loc, varflags & VSQUOTE)) {
595 				amount = loc - expdest;
596 				STADJUST(amount, expdest);
597 				return 1;
598 			}
599 			if ((varflags & VSQUOTE) && *loc == CTLESC)
600 				loc++;
601 		}
602 		return 0;
603 
604 
605 	default:
606 		abort();
607 	}
608 
609 recordleft:
610 	amount = ((str - 1) - (loc - startp)) - expdest;
611 	STADJUST(amount, expdest);
612 	while (loc != str - 1)
613 		*startp++ = *loc++;
614 	return 1;
615 }
616 
617 
618 /*
619  * Expand a variable, and return a pointer to the next character in the
620  * input string.
621  */
622 
623 STATIC char *
624 evalvar(char *p, int flag)
625 {
626 	int subtype;
627 	int varflags;
628 	char *var;
629 	char *val;
630 	int patloc;
631 	int c;
632 	int set;
633 	int special;
634 	int startloc;
635 	int varlen;
636 	int easy;
637 	int quotes = flag & (EXP_FULL | EXP_CASE);
638 
639 	varflags = *p++;
640 	subtype = varflags & VSTYPE;
641 	var = p;
642 	special = 0;
643 	if (! is_name(*p))
644 		special = 1;
645 	p = strchr(p, '=') + 1;
646 again: /* jump here after setting a variable with ${var=text} */
647 	if (special) {
648 		set = varisset(var, varflags & VSNUL);
649 		val = NULL;
650 	} else {
651 		val = bltinlookup(var, 1);
652 		if (val == NULL || ((varflags & VSNUL) && val[0] == '\0')) {
653 			val = NULL;
654 			set = 0;
655 		} else
656 			set = 1;
657 	}
658 	varlen = 0;
659 	startloc = expdest - stackblock();
660 	if (set && subtype != VSPLUS) {
661 		/* insert the value of the variable */
662 		if (special) {
663 			varvalue(var, varflags & VSQUOTE, flag & EXP_FULL);
664 			if (subtype == VSLENGTH) {
665 				varlen = expdest - stackblock() - startloc;
666 				STADJUST(-varlen, expdest);
667 			}
668 		} else {
669 			char const *syntax = (varflags & VSQUOTE) ? DQSYNTAX
670 								  : BASESYNTAX;
671 
672 			if (subtype == VSLENGTH) {
673 				for (;*val; val++)
674 					varlen++;
675 			}
676 			else {
677 				while (*val) {
678 					if (quotes &&
679 					    syntax[(int)*val] == CCTL)
680 						STPUTC(CTLESC, expdest);
681 					STPUTC(*val++, expdest);
682 				}
683 
684 			}
685 		}
686 	}
687 
688 	if (subtype == VSPLUS)
689 		set = ! set;
690 
691 	easy = ((varflags & VSQUOTE) == 0 ||
692 		(*var == '@' && shellparam.nparam != 1));
693 
694 
695 	switch (subtype) {
696 	case VSLENGTH:
697 		expdest = cvtnum(varlen, expdest);
698 		goto record;
699 
700 	case VSNORMAL:
701 		if (!easy)
702 			break;
703 record:
704 		recordregion(startloc, expdest - stackblock(),
705 			     varflags & VSQUOTE);
706 		break;
707 
708 	case VSPLUS:
709 	case VSMINUS:
710 		if (!set) {
711 			argstr(p, flag);
712 			break;
713 		}
714 		if (easy)
715 			goto record;
716 		break;
717 
718 	case VSTRIMLEFT:
719 	case VSTRIMLEFTMAX:
720 	case VSTRIMRIGHT:
721 	case VSTRIMRIGHTMAX:
722 		if (!set)
723 			break;
724 		/*
725 		 * Terminate the string and start recording the pattern
726 		 * right after it
727 		 */
728 		STPUTC('\0', expdest);
729 		patloc = expdest - stackblock();
730 		if (subevalvar(p, NULL, patloc, subtype,
731 			       startloc, varflags) == 0) {
732 			int amount = (expdest - stackblock() - patloc) + 1;
733 			STADJUST(-amount, expdest);
734 		}
735 		/* Remove any recorded regions beyond start of variable */
736 		removerecordregions(startloc);
737 		goto record;
738 
739 	case VSASSIGN:
740 	case VSQUESTION:
741 		if (!set) {
742 			if (subevalvar(p, var, 0, subtype, startloc, varflags)) {
743 				varflags &= ~VSNUL;
744 				/*
745 				 * Remove any recorded regions beyond
746 				 * start of variable
747 				 */
748 				removerecordregions(startloc);
749 				goto again;
750 			}
751 			break;
752 		}
753 		if (easy)
754 			goto record;
755 		break;
756 
757 	default:
758 		abort();
759 	}
760 
761 	if (subtype != VSNORMAL) {	/* skip to end of alternative */
762 		int nesting = 1;
763 		for (;;) {
764 			if ((c = *p++) == CTLESC)
765 				p++;
766 			else if (c == CTLBACKQ || c == (CTLBACKQ|CTLQUOTE)) {
767 				if (set)
768 					argbackq = argbackq->next;
769 			} else if (c == CTLVAR) {
770 				if ((*p++ & VSTYPE) != VSNORMAL)
771 					nesting++;
772 			} else if (c == CTLENDVAR) {
773 				if (--nesting == 0)
774 					break;
775 			}
776 		}
777 	}
778 	return p;
779 }
780 
781 
782 
783 /*
784  * Test whether a specialized variable is set.
785  */
786 
787 STATIC int
788 varisset(char *name, int nulok)
789 {
790 
791 	if (*name == '!')
792 		return backgndpid != -1;
793 	else if (*name == '@' || *name == '*') {
794 		if (*shellparam.p == NULL)
795 			return 0;
796 
797 		if (nulok) {
798 			char **av;
799 
800 			for (av = shellparam.p; *av; av++)
801 				if (**av != '\0')
802 					return 1;
803 			return 0;
804 		}
805 	} else if (is_digit(*name)) {
806 		char *ap;
807 		int num = atoi(name);
808 
809 		if (num > shellparam.nparam)
810 			return 0;
811 
812 		if (num == 0)
813 			ap = arg0;
814 		else
815 			ap = shellparam.p[num - 1];
816 
817 		if (nulok && (ap == NULL || *ap == '\0'))
818 			return 0;
819 	}
820 	return 1;
821 }
822 
823 
824 
825 /*
826  * Add the value of a specialized variable to the stack string.
827  */
828 
829 STATIC void
830 varvalue(char *name, int quoted, int allow_split)
831 {
832 	int num;
833 	char *p;
834 	int i;
835 	extern int oexitstatus;
836 	char sep;
837 	char **ap;
838 	char const *syntax;
839 
840 #define STRTODEST(p) \
841 	do {\
842 	if (allow_split) { \
843 		syntax = quoted? DQSYNTAX : BASESYNTAX; \
844 		while (*p) { \
845 			if (syntax[(int)*p] == CCTL) \
846 				STPUTC(CTLESC, expdest); \
847 			STPUTC(*p++, expdest); \
848 		} \
849 	} else \
850 		while (*p) \
851 			STPUTC(*p++, expdest); \
852 	} while (0)
853 
854 
855 	switch (*name) {
856 	case '$':
857 		num = rootpid;
858 		goto numvar;
859 	case '?':
860 		num = oexitstatus;
861 		goto numvar;
862 	case '#':
863 		num = shellparam.nparam;
864 		goto numvar;
865 	case '!':
866 		num = backgndpid;
867 numvar:
868 		expdest = cvtnum(num, expdest);
869 		break;
870 	case '-':
871 		for (i = 0 ; i < NOPTS ; i++) {
872 			if (optlist[i].val)
873 				STPUTC(optlist[i].letter, expdest);
874 		}
875 		break;
876 	case '@':
877 		if (allow_split && quoted) {
878 			for (ap = shellparam.p ; (p = *ap++) != NULL ; ) {
879 				STRTODEST(p);
880 				if (*ap)
881 					STPUTC('\0', expdest);
882 			}
883 			break;
884 		}
885 		/* fall through */
886 	case '*':
887 		if (ifsset() != 0)
888 			sep = ifsval()[0];
889 		else
890 			sep = ' ';
891 		for (ap = shellparam.p ; (p = *ap++) != NULL ; ) {
892 			STRTODEST(p);
893 			if (*ap && sep)
894 				STPUTC(sep, expdest);
895 		}
896 		break;
897 	case '0':
898 		p = arg0;
899 		STRTODEST(p);
900 		break;
901 	default:
902 		if (is_digit(*name)) {
903 			num = atoi(name);
904 			if (num > 0 && num <= shellparam.nparam) {
905 				p = shellparam.p[num - 1];
906 				STRTODEST(p);
907 			}
908 		}
909 		break;
910 	}
911 }
912 
913 
914 
915 /*
916  * Record the the fact that we have to scan this region of the
917  * string for IFS characters.
918  */
919 
920 STATIC void
921 recordregion(int start, int end, int nulonly)
922 {
923 	struct ifsregion *ifsp;
924 
925 	if (ifslastp == NULL) {
926 		ifsp = &ifsfirst;
927 	} else {
928 		ifsp = (struct ifsregion *)ckmalloc(sizeof (struct ifsregion));
929 		ifslastp->next = ifsp;
930 	}
931 	ifslastp = ifsp;
932 	ifslastp->next = NULL;
933 	ifslastp->begoff = start;
934 	ifslastp->endoff = end;
935 	ifslastp->nulonly = nulonly;
936 }
937 
938 
939 
940 /*
941  * Break the argument string into pieces based upon IFS and add the
942  * strings to the argument list.  The regions of the string to be
943  * searched for IFS characters have been stored by recordregion.
944  */
945 STATIC void
946 ifsbreakup(char *string, struct arglist *arglist)
947 {
948 	struct ifsregion *ifsp;
949 	struct strlist *sp;
950 	char *start;
951 	char *p;
952 	char *q;
953 	char *ifs;
954 	int ifsspc;
955 	int nulonly;
956 
957 
958 	start = string;
959 	ifsspc = 0;
960 	nulonly = 0;
961 	if (ifslastp != NULL) {
962 		ifsp = &ifsfirst;
963 		do {
964 			p = string + ifsp->begoff;
965 			nulonly = ifsp->nulonly;
966 			ifs = nulonly ? nullstr :
967 				( ifsset() ? ifsval() : " \t\n" );
968 			ifsspc = 0;
969 			while (p < string + ifsp->endoff) {
970 				q = p;
971 				if (*p == CTLESC)
972 					p++;
973 				if (strchr(ifs, *p)) {
974 					if (!nulonly)
975 						ifsspc = (strchr(" \t\n", *p) != NULL);
976 					/* Ignore IFS whitespace at start */
977 					if (q == start && ifsspc) {
978 						p++;
979 						start = p;
980 						continue;
981 					}
982 					*q = '\0';
983 					sp = (struct strlist *)stalloc(sizeof *sp);
984 					sp->text = start;
985 					*arglist->lastp = sp;
986 					arglist->lastp = &sp->next;
987 					p++;
988 					if (!nulonly) {
989 						for (;;) {
990 							if (p >= string + ifsp->endoff) {
991 								break;
992 							}
993 							q = p;
994 							if (*p == CTLESC)
995 								p++;
996 							if (strchr(ifs, *p) == NULL ) {
997 								p = q;
998 								break;
999 							} else if (strchr(" \t\n",*p) == NULL) {
1000 								if (ifsspc) {
1001 									p++;
1002 									ifsspc = 0;
1003 								} else {
1004 									p = q;
1005 									break;
1006 								}
1007 							} else
1008 								p++;
1009 						}
1010 					}
1011 					start = p;
1012 				} else
1013 					p++;
1014 			}
1015 		} while ((ifsp = ifsp->next) != NULL);
1016 		if (*start || (!ifsspc && start > string &&
1017 			(nulonly || 1))) {
1018 			sp = (struct strlist *)stalloc(sizeof *sp);
1019 			sp->text = start;
1020 			*arglist->lastp = sp;
1021 			arglist->lastp = &sp->next;
1022 		}
1023 	} else {
1024 		sp = (struct strlist *)stalloc(sizeof *sp);
1025 		sp->text = start;
1026 		*arglist->lastp = sp;
1027 		arglist->lastp = &sp->next;
1028 	}
1029 }
1030 
1031 
1032 
1033 /*
1034  * Expand shell metacharacters.  At this point, the only control characters
1035  * should be escapes.  The results are stored in the list exparg.
1036  */
1037 
1038 char *expdir;
1039 
1040 
1041 STATIC void
1042 expandmeta(struct strlist *str, int flag __unused)
1043 {
1044 	char *p;
1045 	struct strlist **savelastp;
1046 	struct strlist *sp;
1047 	char c;
1048 	/* TODO - EXP_REDIR */
1049 
1050 	while (str) {
1051 		if (fflag)
1052 			goto nometa;
1053 		p = str->text;
1054 		for (;;) {			/* fast check for meta chars */
1055 			if ((c = *p++) == '\0')
1056 				goto nometa;
1057 			if (c == '*' || c == '?' || c == '[' || c == '!')
1058 				break;
1059 		}
1060 		savelastp = exparg.lastp;
1061 		INTOFF;
1062 		if (expdir == NULL) {
1063 			int i = strlen(str->text);
1064 			expdir = ckmalloc(i < 2048 ? 2048 : i); /* XXX */
1065 		}
1066 
1067 		expmeta(expdir, str->text);
1068 		ckfree(expdir);
1069 		expdir = NULL;
1070 		INTON;
1071 		if (exparg.lastp == savelastp) {
1072 			/*
1073 			 * no matches
1074 			 */
1075 nometa:
1076 			*exparg.lastp = str;
1077 			rmescapes(str->text);
1078 			exparg.lastp = &str->next;
1079 		} else {
1080 			*exparg.lastp = NULL;
1081 			*savelastp = sp = expsort(*savelastp);
1082 			while (sp->next != NULL)
1083 				sp = sp->next;
1084 			exparg.lastp = &sp->next;
1085 		}
1086 		str = str->next;
1087 	}
1088 }
1089 
1090 
1091 /*
1092  * Do metacharacter (i.e. *, ?, [...]) expansion.
1093  */
1094 
1095 STATIC void
1096 expmeta(char *enddir, char *name)
1097 {
1098 	char *p;
1099 	char *q;
1100 	char *start;
1101 	char *endname;
1102 	int metaflag;
1103 	struct stat statb;
1104 	DIR *dirp;
1105 	struct dirent *dp;
1106 	int atend;
1107 	int matchdot;
1108 
1109 	metaflag = 0;
1110 	start = name;
1111 	for (p = name ; ; p++) {
1112 		if (*p == '*' || *p == '?')
1113 			metaflag = 1;
1114 		else if (*p == '[') {
1115 			q = p + 1;
1116 			if (*q == '!' || *q == '^')
1117 				q++;
1118 			for (;;) {
1119 				while (*q == CTLQUOTEMARK)
1120 					q++;
1121 				if (*q == CTLESC)
1122 					q++;
1123 				if (*q == '/' || *q == '\0')
1124 					break;
1125 				if (*++q == ']') {
1126 					metaflag = 1;
1127 					break;
1128 				}
1129 			}
1130 		} else if (*p == '!' && p[1] == '!'	&& (p == name || p[-1] == '/')) {
1131 			metaflag = 1;
1132 		} else if (*p == '\0')
1133 			break;
1134 		else if (*p == CTLQUOTEMARK)
1135 			continue;
1136 		else if (*p == CTLESC)
1137 			p++;
1138 		if (*p == '/') {
1139 			if (metaflag)
1140 				break;
1141 			start = p + 1;
1142 		}
1143 	}
1144 	if (metaflag == 0) {	/* we've reached the end of the file name */
1145 		if (enddir != expdir)
1146 			metaflag++;
1147 		for (p = name ; ; p++) {
1148 			if (*p == CTLQUOTEMARK)
1149 				continue;
1150 			if (*p == CTLESC)
1151 				p++;
1152 			*enddir++ = *p;
1153 			if (*p == '\0')
1154 				break;
1155 		}
1156 		if (metaflag == 0 || stat(expdir, &statb) >= 0)
1157 			addfname(expdir);
1158 		return;
1159 	}
1160 	endname = p;
1161 	if (start != name) {
1162 		p = name;
1163 		while (p < start) {
1164 			while (*p == CTLQUOTEMARK)
1165 				p++;
1166 			if (*p == CTLESC)
1167 				p++;
1168 			*enddir++ = *p++;
1169 		}
1170 	}
1171 	if (enddir == expdir) {
1172 		p = ".";
1173 	} else if (enddir == expdir + 1 && *expdir == '/') {
1174 		p = "/";
1175 	} else {
1176 		p = expdir;
1177 		enddir[-1] = '\0';
1178 	}
1179 	if ((dirp = opendir(p)) == NULL)
1180 		return;
1181 	if (enddir != expdir)
1182 		enddir[-1] = '/';
1183 	if (*endname == 0) {
1184 		atend = 1;
1185 	} else {
1186 		atend = 0;
1187 		*endname++ = '\0';
1188 	}
1189 	matchdot = 0;
1190 	p = start;
1191 	while (*p == CTLQUOTEMARK)
1192 		p++;
1193 	if (*p == CTLESC)
1194 		p++;
1195 	if (*p == '.')
1196 		matchdot++;
1197 	while (! int_pending() && (dp = readdir(dirp)) != NULL) {
1198 		if (dp->d_name[0] == '.' && ! matchdot)
1199 			continue;
1200 		if (patmatch(start, dp->d_name, 0)) {
1201 			if (atend) {
1202 				scopy(dp->d_name, enddir);
1203 				addfname(expdir);
1204 			} else {
1205 				char *q;
1206 				for (p = enddir, q = dp->d_name;
1207 				     (*p++ = *q++) != '\0';)
1208 					continue;
1209 				p[-1] = '/';
1210 				expmeta(p, endname);
1211 			}
1212 		}
1213 	}
1214 	closedir(dirp);
1215 	if (! atend)
1216 		endname[-1] = '/';
1217 }
1218 
1219 
1220 /*
1221  * Add a file name to the list.
1222  */
1223 
1224 STATIC void
1225 addfname(char *name)
1226 {
1227 	char *p;
1228 	struct strlist *sp;
1229 
1230 	p = stalloc(strlen(name) + 1);
1231 	scopy(name, p);
1232 	sp = (struct strlist *)stalloc(sizeof *sp);
1233 	sp->text = p;
1234 	*exparg.lastp = sp;
1235 	exparg.lastp = &sp->next;
1236 }
1237 
1238 
1239 /*
1240  * Sort the results of file name expansion.  It calculates the number of
1241  * strings to sort and then calls msort (short for merge sort) to do the
1242  * work.
1243  */
1244 
1245 STATIC struct strlist *
1246 expsort(struct strlist *str)
1247 {
1248 	int len;
1249 	struct strlist *sp;
1250 
1251 	len = 0;
1252 	for (sp = str ; sp ; sp = sp->next)
1253 		len++;
1254 	return msort(str, len);
1255 }
1256 
1257 
1258 STATIC struct strlist *
1259 msort(struct strlist *list, int len)
1260 {
1261 	struct strlist *p, *q = NULL;
1262 	struct strlist **lpp;
1263 	int half;
1264 	int n;
1265 
1266 	if (len <= 1)
1267 		return list;
1268 	half = len >> 1;
1269 	p = list;
1270 	for (n = half ; --n >= 0 ; ) {
1271 		q = p;
1272 		p = p->next;
1273 	}
1274 	q->next = NULL;			/* terminate first half of list */
1275 	q = msort(list, half);		/* sort first half of list */
1276 	p = msort(p, len - half);		/* sort second half */
1277 	lpp = &list;
1278 	for (;;) {
1279 		if (strcmp(p->text, q->text) < 0) {
1280 			*lpp = p;
1281 			lpp = &p->next;
1282 			if ((p = *lpp) == NULL) {
1283 				*lpp = q;
1284 				break;
1285 			}
1286 		} else {
1287 			*lpp = q;
1288 			lpp = &q->next;
1289 			if ((q = *lpp) == NULL) {
1290 				*lpp = p;
1291 				break;
1292 			}
1293 		}
1294 	}
1295 	return list;
1296 }
1297 
1298 
1299 
1300 /*
1301  * Returns true if the pattern matches the string.
1302  */
1303 
1304 int
1305 patmatch(char *pattern, char *string, int squoted)
1306 {
1307 #ifdef notdef
1308 	if (pattern[0] == '!' && pattern[1] == '!')
1309 		return 1 - pmatch(pattern + 2, string);
1310 	else
1311 #endif
1312 		return pmatch(pattern, string, squoted);
1313 }
1314 
1315 
1316 STATIC int
1317 pmatch(char *pattern, char *string, int squoted)
1318 {
1319 	char *p, *q;
1320 	char c;
1321 
1322 	p = pattern;
1323 	q = string;
1324 	for (;;) {
1325 		switch (c = *p++) {
1326 		case '\0':
1327 			goto breakloop;
1328 		case CTLESC:
1329 			if (squoted && *q == CTLESC)
1330 				q++;
1331 			if (*q++ != *p++)
1332 				return 0;
1333 			break;
1334 		case CTLQUOTEMARK:
1335 			continue;
1336 		case '?':
1337 			if (squoted && *q == CTLESC)
1338 				q++;
1339 			if (*q++ == '\0')
1340 				return 0;
1341 			break;
1342 		case '*':
1343 			c = *p;
1344 			while (c == CTLQUOTEMARK || c == '*')
1345 				c = *++p;
1346 			if (c != CTLESC &&  c != CTLQUOTEMARK &&
1347 			    c != '?' && c != '*' && c != '[') {
1348 				while (*q != c) {
1349 					if (squoted && *q == CTLESC &&
1350 					    q[1] == c)
1351 						break;
1352 					if (*q == '\0')
1353 						return 0;
1354 					if (squoted && *q == CTLESC)
1355 						q++;
1356 					q++;
1357 				}
1358 			}
1359 			do {
1360 				if (pmatch(p, q, squoted))
1361 					return 1;
1362 				if (squoted && *q == CTLESC)
1363 					q++;
1364 			} while (*q++ != '\0');
1365 			return 0;
1366 		case '[': {
1367 			char *endp;
1368 			int invert, found;
1369 			char chr;
1370 
1371 			endp = p;
1372 			if (*endp == '!' || *endp == '^')
1373 				endp++;
1374 			for (;;) {
1375 				while (*endp == CTLQUOTEMARK)
1376 					endp++;
1377 				if (*endp == '\0')
1378 					goto dft;		/* no matching ] */
1379 				if (*endp == CTLESC)
1380 					endp++;
1381 				if (*++endp == ']')
1382 					break;
1383 			}
1384 			invert = 0;
1385 			if (*p == '!' || *p == '^') {
1386 				invert++;
1387 				p++;
1388 			}
1389 			found = 0;
1390 			chr = *q++;
1391 			if (squoted && chr == CTLESC)
1392 				chr = *q++;
1393 			if (chr == '\0')
1394 				return 0;
1395 			c = *p++;
1396 			do {
1397 				if (c == CTLQUOTEMARK)
1398 					continue;
1399 				if (c == CTLESC)
1400 					c = *p++;
1401 				if (*p == '-' && p[1] != ']') {
1402 					p++;
1403 					while (*p == CTLQUOTEMARK)
1404 						p++;
1405 					if (*p == CTLESC)
1406 						p++;
1407 					if (   collate_range_cmp(chr, c) >= 0
1408 					    && collate_range_cmp(chr, *p) <= 0
1409 					   )
1410 						found = 1;
1411 					p++;
1412 				} else {
1413 					if (chr == c)
1414 						found = 1;
1415 				}
1416 			} while ((c = *p++) != ']');
1417 			if (found == invert)
1418 				return 0;
1419 			break;
1420 		}
1421 dft:	        default:
1422 			if (squoted && *q == CTLESC)
1423 				q++;
1424 			if (*q++ != c)
1425 				return 0;
1426 			break;
1427 		}
1428 	}
1429 breakloop:
1430 	if (*q != '\0')
1431 		return 0;
1432 	return 1;
1433 }
1434 
1435 
1436 
1437 /*
1438  * Remove any CTLESC characters from a string.
1439  */
1440 
1441 void
1442 rmescapes(char *str)
1443 {
1444 	char *p, *q;
1445 
1446 	p = str;
1447 	while (*p != CTLESC && *p != CTLQUOTEMARK) {
1448 		if (*p++ == '\0')
1449 			return;
1450 	}
1451 	q = p;
1452 	while (*p) {
1453 		if (*p == CTLQUOTEMARK) {
1454 			p++;
1455 			continue;
1456 		}
1457 		if (*p == CTLESC)
1458 			p++;
1459 		*q++ = *p++;
1460 	}
1461 	*q = '\0';
1462 }
1463 
1464 
1465 
1466 /*
1467  * See if a pattern matches in a case statement.
1468  */
1469 
1470 int
1471 casematch(union node *pattern, char *val)
1472 {
1473 	struct stackmark smark;
1474 	int result;
1475 	char *p;
1476 
1477 	setstackmark(&smark);
1478 	argbackq = pattern->narg.backquote;
1479 	STARTSTACKSTR(expdest);
1480 	ifslastp = NULL;
1481 	argstr(pattern->narg.text, EXP_TILDE | EXP_CASE);
1482 	STPUTC('\0', expdest);
1483 	p = grabstackstr(expdest);
1484 	result = patmatch(p, val, 0);
1485 	popstackmark(&smark);
1486 	return result;
1487 }
1488 
1489 /*
1490  * Our own itoa().
1491  */
1492 
1493 STATIC char *
1494 cvtnum(int num, char *buf)
1495 {
1496 	char temp[32];
1497 	int neg = num < 0;
1498 	char *p = temp + 31;
1499 
1500 	temp[31] = '\0';
1501 
1502 	do {
1503 		*--p = num % 10 + '0';
1504 	} while ((num /= 10) != 0);
1505 
1506 	if (neg)
1507 		*--p = '-';
1508 
1509 	while (*p)
1510 		STPUTC(*p++, buf);
1511 	return buf;
1512 }
1513