xref: /freebsd/bin/sh/expand.c (revision d086ded32300bc0f33fb1574d0bcfccfbc60881d)
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 	int nnl;
441 
442 	INTOFF;
443 	saveifs = ifsfirst;
444 	savelastp = ifslastp;
445 	saveargbackq = argbackq;
446 	saveherefd = herefd;
447 	herefd = -1;
448 	p = grabstackstr(dest);
449 	evalbackcmd(cmd, &in);
450 	ungrabstackstr(p, dest);
451 	ifsfirst = saveifs;
452 	ifslastp = savelastp;
453 	argbackq = saveargbackq;
454 	herefd = saveherefd;
455 
456 	p = in.buf;
457 	lastc = '\0';
458 	nnl = 0;
459 	/* Don't copy trailing newlines */
460 	for (;;) {
461 		if (--in.nleft < 0) {
462 			if (in.fd < 0)
463 				break;
464 			while ((i = read(in.fd, buf, sizeof buf)) < 0 && errno == EINTR);
465 			TRACE(("expbackq: read returns %d\n", i));
466 			if (i <= 0)
467 				break;
468 			p = buf;
469 			in.nleft = i - 1;
470 		}
471 		lastc = *p++;
472 		if (lastc != '\0') {
473 			if (quotes && syntax[(int)lastc] == CCTL)
474 				STPUTC(CTLESC, dest);
475 			if (lastc == '\n') {
476 				nnl++;
477 			} else {
478 				while (nnl > 0) {
479 					nnl--;
480 					STPUTC('\n', dest);
481 				}
482 				STPUTC(lastc, dest);
483 			}
484 		}
485 	}
486 
487 	if (in.fd >= 0)
488 		close(in.fd);
489 	if (in.buf)
490 		ckfree(in.buf);
491 	if (in.jp)
492 		exitstatus = waitforjob(in.jp, (int *)NULL);
493 	if (quoted == 0)
494 		recordregion(startloc, dest - stackblock(), 0);
495 	TRACE(("evalbackq: size=%d: \"%.*s\"\n",
496 		(dest - stackblock()) - startloc,
497 		(dest - stackblock()) - startloc,
498 		stackblock() + startloc));
499 	expdest = dest;
500 	INTON;
501 }
502 
503 
504 
505 STATIC int
506 subevalvar(char *p, char *str, int strloc, int subtype, int startloc,
507   int varflags)
508 {
509 	char *startp;
510 	char *loc = NULL;
511 	char *q;
512 	int c = 0;
513 	int saveherefd = herefd;
514 	struct nodelist *saveargbackq = argbackq;
515 	int amount;
516 
517 	herefd = -1;
518 	argstr(p, 0);
519 	STACKSTRNUL(expdest);
520 	herefd = saveherefd;
521 	argbackq = saveargbackq;
522 	startp = stackblock() + startloc;
523 	if (str == NULL)
524 	    str = stackblock() + strloc;
525 
526 	switch (subtype) {
527 	case VSASSIGN:
528 		setvar(str, startp, 0);
529 		amount = startp - expdest;
530 		STADJUST(amount, expdest);
531 		varflags &= ~VSNUL;
532 		if (c != 0)
533 			*loc = c;
534 		return 1;
535 
536 	case VSQUESTION:
537 		if (*p != CTLENDVAR) {
538 			outfmt(&errout, "%s\n", startp);
539 			error((char *)NULL);
540 		}
541 		error("%.*s: parameter %snot set", (int)(p - str - 1),
542 		      str, (varflags & VSNUL) ? "null or "
543 					      : nullstr);
544 		return 0;
545 
546 	case VSTRIMLEFT:
547 		for (loc = startp; loc < str; loc++) {
548 			c = *loc;
549 			*loc = '\0';
550 			if (patmatch(str, startp, varflags & VSQUOTE)) {
551 				*loc = c;
552 				goto recordleft;
553 			}
554 			*loc = c;
555 			if ((varflags & VSQUOTE) && *loc == CTLESC)
556 				loc++;
557 		}
558 		return 0;
559 
560 	case VSTRIMLEFTMAX:
561 		for (loc = str - 1; loc >= startp;) {
562 			c = *loc;
563 			*loc = '\0';
564 			if (patmatch(str, startp, varflags & VSQUOTE)) {
565 				*loc = c;
566 				goto recordleft;
567 			}
568 			*loc = c;
569 			loc--;
570 			if ((varflags & VSQUOTE) && loc > startp &&
571 			    *(loc - 1) == CTLESC) {
572 				for (q = startp; q < loc; q++)
573 					if (*q == CTLESC)
574 						q++;
575 				if (q > loc)
576 					loc--;
577 			}
578 		}
579 		return 0;
580 
581 	case VSTRIMRIGHT:
582 		for (loc = str - 1; loc >= startp;) {
583 			if (patmatch(str, loc, varflags & VSQUOTE)) {
584 				amount = loc - expdest;
585 				STADJUST(amount, expdest);
586 				return 1;
587 			}
588 			loc--;
589 			if ((varflags & VSQUOTE) && loc > startp &&
590 			    *(loc - 1) == CTLESC) {
591 				for (q = startp; q < loc; q++)
592 					if (*q == CTLESC)
593 						q++;
594 				if (q > loc)
595 					loc--;
596 			}
597 		}
598 		return 0;
599 
600 	case VSTRIMRIGHTMAX:
601 		for (loc = startp; loc < str - 1; loc++) {
602 			if (patmatch(str, loc, varflags & VSQUOTE)) {
603 				amount = loc - expdest;
604 				STADJUST(amount, expdest);
605 				return 1;
606 			}
607 			if ((varflags & VSQUOTE) && *loc == CTLESC)
608 				loc++;
609 		}
610 		return 0;
611 
612 
613 	default:
614 		abort();
615 	}
616 
617 recordleft:
618 	amount = ((str - 1) - (loc - startp)) - expdest;
619 	STADJUST(amount, expdest);
620 	while (loc != str - 1)
621 		*startp++ = *loc++;
622 	return 1;
623 }
624 
625 
626 /*
627  * Expand a variable, and return a pointer to the next character in the
628  * input string.
629  */
630 
631 STATIC char *
632 evalvar(char *p, int flag)
633 {
634 	int subtype;
635 	int varflags;
636 	char *var;
637 	char *val;
638 	int patloc;
639 	int c;
640 	int set;
641 	int special;
642 	int startloc;
643 	int varlen;
644 	int easy;
645 	int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR);
646 
647 	varflags = *p++;
648 	subtype = varflags & VSTYPE;
649 	var = p;
650 	special = 0;
651 	if (! is_name(*p))
652 		special = 1;
653 	p = strchr(p, '=') + 1;
654 again: /* jump here after setting a variable with ${var=text} */
655 	if (special) {
656 		set = varisset(var, varflags & VSNUL);
657 		val = NULL;
658 	} else {
659 		val = bltinlookup(var, 1);
660 		if (val == NULL || ((varflags & VSNUL) && val[0] == '\0')) {
661 			val = NULL;
662 			set = 0;
663 		} else
664 			set = 1;
665 	}
666 	varlen = 0;
667 	startloc = expdest - stackblock();
668 	if (!set && uflag) {
669 		switch (subtype) {
670 		case VSNORMAL:
671 		case VSTRIMLEFT:
672 		case VSTRIMLEFTMAX:
673 		case VSTRIMRIGHT:
674 		case VSTRIMRIGHTMAX:
675 		case VSLENGTH:
676 			error("%.*s: parameter not set", (int)(p - var - 1),
677 			    var);
678 		}
679 	}
680 	if (set && subtype != VSPLUS) {
681 		/* insert the value of the variable */
682 		if (special) {
683 			varvalue(var, varflags & VSQUOTE, flag & EXP_FULL);
684 			if (subtype == VSLENGTH) {
685 				varlen = expdest - stackblock() - startloc;
686 				STADJUST(-varlen, expdest);
687 			}
688 		} else {
689 			char const *syntax = (varflags & VSQUOTE) ? DQSYNTAX
690 								  : BASESYNTAX;
691 
692 			if (subtype == VSLENGTH) {
693 				for (;*val; val++)
694 					varlen++;
695 			}
696 			else {
697 				while (*val) {
698 					if (quotes &&
699 					    syntax[(int)*val] == CCTL)
700 						STPUTC(CTLESC, expdest);
701 					STPUTC(*val++, expdest);
702 				}
703 
704 			}
705 		}
706 	}
707 
708 	if (subtype == VSPLUS)
709 		set = ! set;
710 
711 	easy = ((varflags & VSQUOTE) == 0 ||
712 		(*var == '@' && shellparam.nparam != 1));
713 
714 
715 	switch (subtype) {
716 	case VSLENGTH:
717 		expdest = cvtnum(varlen, expdest);
718 		goto record;
719 
720 	case VSNORMAL:
721 		if (!easy)
722 			break;
723 record:
724 		recordregion(startloc, expdest - stackblock(),
725 			     varflags & VSQUOTE);
726 		break;
727 
728 	case VSPLUS:
729 	case VSMINUS:
730 		if (!set) {
731 			argstr(p, flag);
732 			break;
733 		}
734 		if (easy)
735 			goto record;
736 		break;
737 
738 	case VSTRIMLEFT:
739 	case VSTRIMLEFTMAX:
740 	case VSTRIMRIGHT:
741 	case VSTRIMRIGHTMAX:
742 		if (!set)
743 			break;
744 		/*
745 		 * Terminate the string and start recording the pattern
746 		 * right after it
747 		 */
748 		STPUTC('\0', expdest);
749 		patloc = expdest - stackblock();
750 		if (subevalvar(p, NULL, patloc, subtype,
751 			       startloc, varflags) == 0) {
752 			int amount = (expdest - stackblock() - patloc) + 1;
753 			STADJUST(-amount, expdest);
754 		}
755 		/* Remove any recorded regions beyond start of variable */
756 		removerecordregions(startloc);
757 		goto record;
758 
759 	case VSASSIGN:
760 	case VSQUESTION:
761 		if (!set) {
762 			if (subevalvar(p, var, 0, subtype, startloc, varflags)) {
763 				varflags &= ~VSNUL;
764 				/*
765 				 * Remove any recorded regions beyond
766 				 * start of variable
767 				 */
768 				removerecordregions(startloc);
769 				goto again;
770 			}
771 			break;
772 		}
773 		if (easy)
774 			goto record;
775 		break;
776 
777 	default:
778 		abort();
779 	}
780 
781 	if (subtype != VSNORMAL) {	/* skip to end of alternative */
782 		int nesting = 1;
783 		for (;;) {
784 			if ((c = *p++) == CTLESC)
785 				p++;
786 			else if (c == CTLBACKQ || c == (CTLBACKQ|CTLQUOTE)) {
787 				if (set)
788 					argbackq = argbackq->next;
789 			} else if (c == CTLVAR) {
790 				if ((*p++ & VSTYPE) != VSNORMAL)
791 					nesting++;
792 			} else if (c == CTLENDVAR) {
793 				if (--nesting == 0)
794 					break;
795 			}
796 		}
797 	}
798 	return p;
799 }
800 
801 
802 
803 /*
804  * Test whether a specialized variable is set.
805  */
806 
807 STATIC int
808 varisset(char *name, int nulok)
809 {
810 
811 	if (*name == '!')
812 		return backgndpid != -1;
813 	else if (*name == '@' || *name == '*') {
814 		if (*shellparam.p == NULL)
815 			return 0;
816 
817 		if (nulok) {
818 			char **av;
819 
820 			for (av = shellparam.p; *av; av++)
821 				if (**av != '\0')
822 					return 1;
823 			return 0;
824 		}
825 	} else if (is_digit(*name)) {
826 		char *ap;
827 		int num = atoi(name);
828 
829 		if (num > shellparam.nparam)
830 			return 0;
831 
832 		if (num == 0)
833 			ap = arg0;
834 		else
835 			ap = shellparam.p[num - 1];
836 
837 		if (nulok && (ap == NULL || *ap == '\0'))
838 			return 0;
839 	}
840 	return 1;
841 }
842 
843 
844 
845 /*
846  * Add the value of a specialized variable to the stack string.
847  */
848 
849 STATIC void
850 varvalue(char *name, int quoted, int allow_split)
851 {
852 	int num;
853 	char *p;
854 	int i;
855 	extern int oexitstatus;
856 	char sep;
857 	char **ap;
858 	char const *syntax;
859 
860 #define STRTODEST(p) \
861 	do {\
862 	if (allow_split) { \
863 		syntax = quoted? DQSYNTAX : BASESYNTAX; \
864 		while (*p) { \
865 			if (syntax[(int)*p] == CCTL) \
866 				STPUTC(CTLESC, expdest); \
867 			STPUTC(*p++, expdest); \
868 		} \
869 	} else \
870 		while (*p) \
871 			STPUTC(*p++, expdest); \
872 	} while (0)
873 
874 
875 	switch (*name) {
876 	case '$':
877 		num = rootpid;
878 		goto numvar;
879 	case '?':
880 		num = oexitstatus;
881 		goto numvar;
882 	case '#':
883 		num = shellparam.nparam;
884 		goto numvar;
885 	case '!':
886 		num = backgndpid;
887 numvar:
888 		expdest = cvtnum(num, expdest);
889 		break;
890 	case '-':
891 		for (i = 0 ; i < NOPTS ; i++) {
892 			if (optlist[i].val)
893 				STPUTC(optlist[i].letter, expdest);
894 		}
895 		break;
896 	case '@':
897 		if (allow_split && quoted) {
898 			for (ap = shellparam.p ; (p = *ap++) != NULL ; ) {
899 				STRTODEST(p);
900 				if (*ap)
901 					STPUTC('\0', expdest);
902 			}
903 			break;
904 		}
905 		/* FALLTHROUGH */
906 	case '*':
907 		if (ifsset() != 0)
908 			sep = ifsval()[0];
909 		else
910 			sep = ' ';
911 		for (ap = shellparam.p ; (p = *ap++) != NULL ; ) {
912 			STRTODEST(p);
913 			if (*ap && sep)
914 				STPUTC(sep, expdest);
915 		}
916 		break;
917 	case '0':
918 		p = arg0;
919 		STRTODEST(p);
920 		break;
921 	default:
922 		if (is_digit(*name)) {
923 			num = atoi(name);
924 			if (num > 0 && num <= shellparam.nparam) {
925 				p = shellparam.p[num - 1];
926 				STRTODEST(p);
927 			}
928 		}
929 		break;
930 	}
931 }
932 
933 
934 
935 /*
936  * Record the the fact that we have to scan this region of the
937  * string for IFS characters.
938  */
939 
940 STATIC void
941 recordregion(int start, int end, int nulonly)
942 {
943 	struct ifsregion *ifsp;
944 
945 	if (ifslastp == NULL) {
946 		ifsp = &ifsfirst;
947 	} else {
948 		ifsp = (struct ifsregion *)ckmalloc(sizeof (struct ifsregion));
949 		ifslastp->next = ifsp;
950 	}
951 	ifslastp = ifsp;
952 	ifslastp->next = NULL;
953 	ifslastp->begoff = start;
954 	ifslastp->endoff = end;
955 	ifslastp->nulonly = nulonly;
956 }
957 
958 
959 
960 /*
961  * Break the argument string into pieces based upon IFS and add the
962  * strings to the argument list.  The regions of the string to be
963  * searched for IFS characters have been stored by recordregion.
964  */
965 STATIC void
966 ifsbreakup(char *string, struct arglist *arglist)
967 {
968 	struct ifsregion *ifsp;
969 	struct strlist *sp;
970 	char *start;
971 	char *p;
972 	char *q;
973 	char *ifs;
974 	int ifsspc;
975 	int nulonly;
976 
977 
978 	start = string;
979 	ifsspc = 0;
980 	nulonly = 0;
981 	if (ifslastp != NULL) {
982 		ifsp = &ifsfirst;
983 		do {
984 			p = string + ifsp->begoff;
985 			nulonly = ifsp->nulonly;
986 			ifs = nulonly ? nullstr :
987 				( ifsset() ? ifsval() : " \t\n" );
988 			ifsspc = 0;
989 			while (p < string + ifsp->endoff) {
990 				q = p;
991 				if (*p == CTLESC)
992 					p++;
993 				if (strchr(ifs, *p)) {
994 					if (!nulonly)
995 						ifsspc = (strchr(" \t\n", *p) != NULL);
996 					/* Ignore IFS whitespace at start */
997 					if (q == start && ifsspc) {
998 						p++;
999 						start = p;
1000 						continue;
1001 					}
1002 					*q = '\0';
1003 					sp = (struct strlist *)stalloc(sizeof *sp);
1004 					sp->text = start;
1005 					*arglist->lastp = sp;
1006 					arglist->lastp = &sp->next;
1007 					p++;
1008 					if (!nulonly) {
1009 						for (;;) {
1010 							if (p >= string + ifsp->endoff) {
1011 								break;
1012 							}
1013 							q = p;
1014 							if (*p == CTLESC)
1015 								p++;
1016 							if (strchr(ifs, *p) == NULL ) {
1017 								p = q;
1018 								break;
1019 							} else if (strchr(" \t\n",*p) == NULL) {
1020 								if (ifsspc) {
1021 									p++;
1022 									ifsspc = 0;
1023 								} else {
1024 									p = q;
1025 									break;
1026 								}
1027 							} else
1028 								p++;
1029 						}
1030 					}
1031 					start = p;
1032 				} else
1033 					p++;
1034 			}
1035 		} while ((ifsp = ifsp->next) != NULL);
1036 		if (*start || (!ifsspc && start > string &&
1037 			(nulonly || 1))) {
1038 			sp = (struct strlist *)stalloc(sizeof *sp);
1039 			sp->text = start;
1040 			*arglist->lastp = sp;
1041 			arglist->lastp = &sp->next;
1042 		}
1043 	} else {
1044 		sp = (struct strlist *)stalloc(sizeof *sp);
1045 		sp->text = start;
1046 		*arglist->lastp = sp;
1047 		arglist->lastp = &sp->next;
1048 	}
1049 }
1050 
1051 
1052 
1053 /*
1054  * Expand shell metacharacters.  At this point, the only control characters
1055  * should be escapes.  The results are stored in the list exparg.
1056  */
1057 
1058 char *expdir;
1059 
1060 
1061 STATIC void
1062 expandmeta(struct strlist *str, int flag __unused)
1063 {
1064 	char *p;
1065 	struct strlist **savelastp;
1066 	struct strlist *sp;
1067 	char c;
1068 	/* TODO - EXP_REDIR */
1069 
1070 	while (str) {
1071 		if (fflag)
1072 			goto nometa;
1073 		p = str->text;
1074 		for (;;) {			/* fast check for meta chars */
1075 			if ((c = *p++) == '\0')
1076 				goto nometa;
1077 			if (c == '*' || c == '?' || c == '[' || c == '!')
1078 				break;
1079 		}
1080 		savelastp = exparg.lastp;
1081 		INTOFF;
1082 		if (expdir == NULL) {
1083 			int i = strlen(str->text);
1084 			expdir = ckmalloc(i < 2048 ? 2048 : i); /* XXX */
1085 		}
1086 
1087 		expmeta(expdir, str->text);
1088 		ckfree(expdir);
1089 		expdir = NULL;
1090 		INTON;
1091 		if (exparg.lastp == savelastp) {
1092 			/*
1093 			 * no matches
1094 			 */
1095 nometa:
1096 			*exparg.lastp = str;
1097 			rmescapes(str->text);
1098 			exparg.lastp = &str->next;
1099 		} else {
1100 			*exparg.lastp = NULL;
1101 			*savelastp = sp = expsort(*savelastp);
1102 			while (sp->next != NULL)
1103 				sp = sp->next;
1104 			exparg.lastp = &sp->next;
1105 		}
1106 		str = str->next;
1107 	}
1108 }
1109 
1110 
1111 /*
1112  * Do metacharacter (i.e. *, ?, [...]) expansion.
1113  */
1114 
1115 STATIC void
1116 expmeta(char *enddir, char *name)
1117 {
1118 	char *p;
1119 	char *q;
1120 	char *start;
1121 	char *endname;
1122 	int metaflag;
1123 	struct stat statb;
1124 	DIR *dirp;
1125 	struct dirent *dp;
1126 	int atend;
1127 	int matchdot;
1128 
1129 	metaflag = 0;
1130 	start = name;
1131 	for (p = name ; ; p++) {
1132 		if (*p == '*' || *p == '?')
1133 			metaflag = 1;
1134 		else if (*p == '[') {
1135 			q = p + 1;
1136 			if (*q == '!' || *q == '^')
1137 				q++;
1138 			for (;;) {
1139 				while (*q == CTLQUOTEMARK)
1140 					q++;
1141 				if (*q == CTLESC)
1142 					q++;
1143 				if (*q == '/' || *q == '\0')
1144 					break;
1145 				if (*++q == ']') {
1146 					metaflag = 1;
1147 					break;
1148 				}
1149 			}
1150 		} else if (*p == '!' && p[1] == '!'	&& (p == name || p[-1] == '/')) {
1151 			metaflag = 1;
1152 		} else if (*p == '\0')
1153 			break;
1154 		else if (*p == CTLQUOTEMARK)
1155 			continue;
1156 		else if (*p == CTLESC)
1157 			p++;
1158 		if (*p == '/') {
1159 			if (metaflag)
1160 				break;
1161 			start = p + 1;
1162 		}
1163 	}
1164 	if (metaflag == 0) {	/* we've reached the end of the file name */
1165 		if (enddir != expdir)
1166 			metaflag++;
1167 		for (p = name ; ; p++) {
1168 			if (*p == CTLQUOTEMARK)
1169 				continue;
1170 			if (*p == CTLESC)
1171 				p++;
1172 			*enddir++ = *p;
1173 			if (*p == '\0')
1174 				break;
1175 		}
1176 		if (metaflag == 0 || stat(expdir, &statb) >= 0)
1177 			addfname(expdir);
1178 		return;
1179 	}
1180 	endname = p;
1181 	if (start != name) {
1182 		p = name;
1183 		while (p < start) {
1184 			while (*p == CTLQUOTEMARK)
1185 				p++;
1186 			if (*p == CTLESC)
1187 				p++;
1188 			*enddir++ = *p++;
1189 		}
1190 	}
1191 	if (enddir == expdir) {
1192 		p = ".";
1193 	} else if (enddir == expdir + 1 && *expdir == '/') {
1194 		p = "/";
1195 	} else {
1196 		p = expdir;
1197 		enddir[-1] = '\0';
1198 	}
1199 	if ((dirp = opendir(p)) == NULL)
1200 		return;
1201 	if (enddir != expdir)
1202 		enddir[-1] = '/';
1203 	if (*endname == 0) {
1204 		atend = 1;
1205 	} else {
1206 		atend = 0;
1207 		*endname++ = '\0';
1208 	}
1209 	matchdot = 0;
1210 	p = start;
1211 	while (*p == CTLQUOTEMARK)
1212 		p++;
1213 	if (*p == CTLESC)
1214 		p++;
1215 	if (*p == '.')
1216 		matchdot++;
1217 	while (! int_pending() && (dp = readdir(dirp)) != NULL) {
1218 		if (dp->d_name[0] == '.' && ! matchdot)
1219 			continue;
1220 		if (patmatch(start, dp->d_name, 0)) {
1221 			if (atend) {
1222 				scopy(dp->d_name, enddir);
1223 				addfname(expdir);
1224 			} else {
1225 				char *q;
1226 				for (p = enddir, q = dp->d_name;
1227 				     (*p++ = *q++) != '\0';)
1228 					continue;
1229 				p[-1] = '/';
1230 				expmeta(p, endname);
1231 			}
1232 		}
1233 	}
1234 	closedir(dirp);
1235 	if (! atend)
1236 		endname[-1] = '/';
1237 }
1238 
1239 
1240 /*
1241  * Add a file name to the list.
1242  */
1243 
1244 STATIC void
1245 addfname(char *name)
1246 {
1247 	char *p;
1248 	struct strlist *sp;
1249 
1250 	p = stalloc(strlen(name) + 1);
1251 	scopy(name, p);
1252 	sp = (struct strlist *)stalloc(sizeof *sp);
1253 	sp->text = p;
1254 	*exparg.lastp = sp;
1255 	exparg.lastp = &sp->next;
1256 }
1257 
1258 
1259 /*
1260  * Sort the results of file name expansion.  It calculates the number of
1261  * strings to sort and then calls msort (short for merge sort) to do the
1262  * work.
1263  */
1264 
1265 STATIC struct strlist *
1266 expsort(struct strlist *str)
1267 {
1268 	int len;
1269 	struct strlist *sp;
1270 
1271 	len = 0;
1272 	for (sp = str ; sp ; sp = sp->next)
1273 		len++;
1274 	return msort(str, len);
1275 }
1276 
1277 
1278 STATIC struct strlist *
1279 msort(struct strlist *list, int len)
1280 {
1281 	struct strlist *p, *q = NULL;
1282 	struct strlist **lpp;
1283 	int half;
1284 	int n;
1285 
1286 	if (len <= 1)
1287 		return list;
1288 	half = len >> 1;
1289 	p = list;
1290 	for (n = half ; --n >= 0 ; ) {
1291 		q = p;
1292 		p = p->next;
1293 	}
1294 	q->next = NULL;			/* terminate first half of list */
1295 	q = msort(list, half);		/* sort first half of list */
1296 	p = msort(p, len - half);		/* sort second half */
1297 	lpp = &list;
1298 	for (;;) {
1299 		if (strcmp(p->text, q->text) < 0) {
1300 			*lpp = p;
1301 			lpp = &p->next;
1302 			if ((p = *lpp) == NULL) {
1303 				*lpp = q;
1304 				break;
1305 			}
1306 		} else {
1307 			*lpp = q;
1308 			lpp = &q->next;
1309 			if ((q = *lpp) == NULL) {
1310 				*lpp = p;
1311 				break;
1312 			}
1313 		}
1314 	}
1315 	return list;
1316 }
1317 
1318 
1319 
1320 /*
1321  * Returns true if the pattern matches the string.
1322  */
1323 
1324 int
1325 patmatch(char *pattern, char *string, int squoted)
1326 {
1327 #ifdef notdef
1328 	if (pattern[0] == '!' && pattern[1] == '!')
1329 		return 1 - pmatch(pattern + 2, string);
1330 	else
1331 #endif
1332 		return pmatch(pattern, string, squoted);
1333 }
1334 
1335 
1336 STATIC int
1337 pmatch(char *pattern, char *string, int squoted)
1338 {
1339 	char *p, *q;
1340 	char c;
1341 
1342 	p = pattern;
1343 	q = string;
1344 	for (;;) {
1345 		switch (c = *p++) {
1346 		case '\0':
1347 			goto breakloop;
1348 		case CTLESC:
1349 			if (squoted && *q == CTLESC)
1350 				q++;
1351 			if (*q++ != *p++)
1352 				return 0;
1353 			break;
1354 		case CTLQUOTEMARK:
1355 			continue;
1356 		case '?':
1357 			if (squoted && *q == CTLESC)
1358 				q++;
1359 			if (*q++ == '\0')
1360 				return 0;
1361 			break;
1362 		case '*':
1363 			c = *p;
1364 			while (c == CTLQUOTEMARK || c == '*')
1365 				c = *++p;
1366 			if (c != CTLESC &&  c != CTLQUOTEMARK &&
1367 			    c != '?' && c != '*' && c != '[') {
1368 				while (*q != c) {
1369 					if (squoted && *q == CTLESC &&
1370 					    q[1] == c)
1371 						break;
1372 					if (*q == '\0')
1373 						return 0;
1374 					if (squoted && *q == CTLESC)
1375 						q++;
1376 					q++;
1377 				}
1378 			}
1379 			do {
1380 				if (pmatch(p, q, squoted))
1381 					return 1;
1382 				if (squoted && *q == CTLESC)
1383 					q++;
1384 			} while (*q++ != '\0');
1385 			return 0;
1386 		case '[': {
1387 			char *endp;
1388 			int invert, found;
1389 			char chr;
1390 
1391 			endp = p;
1392 			if (*endp == '!' || *endp == '^')
1393 				endp++;
1394 			for (;;) {
1395 				while (*endp == CTLQUOTEMARK)
1396 					endp++;
1397 				if (*endp == '\0')
1398 					goto dft;		/* no matching ] */
1399 				if (*endp == CTLESC)
1400 					endp++;
1401 				if (*++endp == ']')
1402 					break;
1403 			}
1404 			invert = 0;
1405 			if (*p == '!' || *p == '^') {
1406 				invert++;
1407 				p++;
1408 			}
1409 			found = 0;
1410 			chr = *q++;
1411 			if (squoted && chr == CTLESC)
1412 				chr = *q++;
1413 			if (chr == '\0')
1414 				return 0;
1415 			c = *p++;
1416 			do {
1417 				if (c == CTLQUOTEMARK)
1418 					continue;
1419 				if (c == CTLESC)
1420 					c = *p++;
1421 				if (*p == '-' && p[1] != ']') {
1422 					p++;
1423 					while (*p == CTLQUOTEMARK)
1424 						p++;
1425 					if (*p == CTLESC)
1426 						p++;
1427 					if (   collate_range_cmp(chr, c) >= 0
1428 					    && collate_range_cmp(chr, *p) <= 0
1429 					   )
1430 						found = 1;
1431 					p++;
1432 				} else {
1433 					if (chr == c)
1434 						found = 1;
1435 				}
1436 			} while ((c = *p++) != ']');
1437 			if (found == invert)
1438 				return 0;
1439 			break;
1440 		}
1441 dft:	        default:
1442 			if (squoted && *q == CTLESC)
1443 				q++;
1444 			if (*q++ != c)
1445 				return 0;
1446 			break;
1447 		}
1448 	}
1449 breakloop:
1450 	if (*q != '\0')
1451 		return 0;
1452 	return 1;
1453 }
1454 
1455 
1456 
1457 /*
1458  * Remove any CTLESC characters from a string.
1459  */
1460 
1461 void
1462 rmescapes(char *str)
1463 {
1464 	char *p, *q;
1465 
1466 	p = str;
1467 	while (*p != CTLESC && *p != CTLQUOTEMARK) {
1468 		if (*p++ == '\0')
1469 			return;
1470 	}
1471 	q = p;
1472 	while (*p) {
1473 		if (*p == CTLQUOTEMARK) {
1474 			p++;
1475 			continue;
1476 		}
1477 		if (*p == CTLESC)
1478 			p++;
1479 		*q++ = *p++;
1480 	}
1481 	*q = '\0';
1482 }
1483 
1484 
1485 
1486 /*
1487  * See if a pattern matches in a case statement.
1488  */
1489 
1490 int
1491 casematch(union node *pattern, char *val)
1492 {
1493 	struct stackmark smark;
1494 	int result;
1495 	char *p;
1496 
1497 	setstackmark(&smark);
1498 	argbackq = pattern->narg.backquote;
1499 	STARTSTACKSTR(expdest);
1500 	ifslastp = NULL;
1501 	argstr(pattern->narg.text, EXP_TILDE | EXP_CASE);
1502 	STPUTC('\0', expdest);
1503 	p = grabstackstr(expdest);
1504 	result = patmatch(p, val, 0);
1505 	popstackmark(&smark);
1506 	return result;
1507 }
1508 
1509 /*
1510  * Our own itoa().
1511  */
1512 
1513 STATIC char *
1514 cvtnum(int num, char *buf)
1515 {
1516 	char temp[32];
1517 	int neg = num < 0;
1518 	char *p = temp + 31;
1519 
1520 	temp[31] = '\0';
1521 
1522 	do {
1523 		*--p = num % 10 + '0';
1524 	} while ((num /= 10) != 0);
1525 
1526 	if (neg)
1527 		*--p = '-';
1528 
1529 	while (*p)
1530 		STPUTC(*p++, buf);
1531 	return buf;
1532 }
1533 
1534 /*
1535  * Do most of the work for wordexp(3).
1536  */
1537 
1538 int
1539 wordexpcmd(int argc, char **argv)
1540 {
1541 	size_t len;
1542 	int i;
1543 
1544 	out1fmt("%08x", argc - 1);
1545 	for (i = 1, len = 0; i < argc; i++)
1546 		len += strlen(argv[i]);
1547 	out1fmt("%08x", (int)len);
1548 	for (i = 1; i < argc; i++) {
1549 		out1str(argv[i]);
1550 		out1c('\0');
1551 	}
1552         return (0);
1553 }
1554