xref: /netbsd-src/bin/ksh/eval.c (revision 23c8222edbfb0f0932d88a8351d3a0cf817dfb9e)
1 /*	$NetBSD: eval.c,v 1.5 2004/07/07 19:20:09 mycroft Exp $	*/
2 
3 /*
4  * Expansion - quoting, separation, substitution, globbing
5  */
6 #include <sys/cdefs.h>
7 
8 #ifndef lint
9 __RCSID("$NetBSD: eval.c,v 1.5 2004/07/07 19:20:09 mycroft Exp $");
10 #endif
11 
12 
13 #include "sh.h"
14 #include <pwd.h>
15 #include "ksh_dir.h"
16 #include "ksh_stat.h"
17 
18 /*
19  * string expansion
20  *
21  * first pass: quoting, IFS separation, ~, ${}, $() and $(()) substitution.
22  * second pass: alternation ({,}), filename expansion (*?[]).
23  */
24 
25 /* expansion generator state */
26 typedef struct Expand {
27 	/* int  type; */	/* see expand() */
28 	const char *str;	/* string */
29 	union {
30 		const char **strv;/* string[] */
31 		struct shf *shf;/* file */
32 	} u;			/* source */
33 	struct tbl *var;	/* variable in ${var..} */
34 	short	split;		/* split "$@" / call waitlast $() */
35 } Expand;
36 
37 #define	XBASE		0	/* scanning original */
38 #define	XSUB		1	/* expanding ${} string */
39 #define	XARGSEP		2	/* ifs0 between "$*" */
40 #define	XARG		3	/* expanding $*, $@ */
41 #define	XCOM		4	/* expanding $() */
42 #define XNULLSUB	5	/* "$@" when $# is 0 (don't generate word) */
43 
44 /* States used for field splitting */
45 #define IFS_WORD	0	/* word has chars (or quotes) */
46 #define IFS_WS		1	/* have seen IFS white-space */
47 #define IFS_NWS		2	/* have seen IFS non-white-space */
48 
49 static	int	varsub ARGS((Expand *xp, char *sp, char *word, int *stypep, int *slenp));
50 static	int	comsub ARGS((Expand *xp, char *cp));
51 static	char   *trimsub ARGS((char *str, char *pat, int how));
52 static	void	glob ARGS((char *cp, XPtrV *wp, int markdirs));
53 static	void	globit ARGS((XString *xs, char **xpp, char *sp, XPtrV *wp,
54 			     int check));
55 static char	*maybe_expand_tilde ARGS((char *p, XString *dsp, char **dpp,
56 					  int isassign));
57 static	char   *tilde ARGS((char *acp));
58 static	char   *homedir ARGS((char *name));
59 #ifdef BRACE_EXPAND
60 static void	alt_expand ARGS((XPtrV *wp, char *start, char *exp_start,
61 				 char *end, int fdo));
62 #endif
63 
64 /* compile and expand word */
65 char *
66 substitute(cp, f)
67 	const char *cp;
68 	int f;
69 {
70 	struct source *s, *sold;
71 
72 	sold = source;
73 	s = pushs(SWSTR, ATEMP);
74 	s->start = s->str = cp;
75 	source = s;
76 	if (yylex(ONEWORD) != LWORD)
77 		internal_errorf(1, "substitute");
78 	source = sold;
79 	afree(s, ATEMP);
80 	return evalstr(yylval.cp, f);
81 }
82 
83 /*
84  * expand arg-list
85  */
86 char **
87 eval(ap, f)
88 	register char **ap;
89 	int f;
90 {
91 	XPtrV w;
92 
93 	if (*ap == NULL)
94 		return ap;
95 	XPinit(w, 32);
96 	XPput(w, NULL);		/* space for shell name */
97 #ifdef	SHARPBANG
98 	XPput(w, NULL);		/* and space for one arg */
99 #endif
100 	while (*ap != NULL)
101 		expand(*ap++, &w, f);
102 	XPput(w, NULL);
103 #ifdef	SHARPBANG
104 	return (char **) XPclose(w) + 2;
105 #else
106 	return (char **) XPclose(w) + 1;
107 #endif
108 }
109 
110 /*
111  * expand string
112  */
113 char *
114 evalstr(cp, f)
115 	char *cp;
116 	int f;
117 {
118 	XPtrV w;
119 
120 	XPinit(w, 1);
121 	expand(cp, &w, f);
122 	cp = (XPsize(w) == 0) ? null : (char*) *XPptrv(w);
123 	XPfree(w);
124 	return cp;
125 }
126 
127 /*
128  * expand string - return only one component
129  * used from iosetup to expand redirection files
130  */
131 char *
132 evalonestr(cp, f)
133 	register char *cp;
134 	int f;
135 {
136 	XPtrV w;
137 
138 	XPinit(w, 1);
139 	expand(cp, &w, f);
140 	switch (XPsize(w)) {
141 	case 0:
142 		cp = null;
143 		break;
144 	case 1:
145 		cp = (char*) *XPptrv(w);
146 		break;
147 	default:
148 		cp = evalstr(cp, f&~DOGLOB);
149 		break;
150 	}
151 	XPfree(w);
152 	return cp;
153 }
154 
155 /* for nested substitution: ${var:=$var2} */
156 typedef struct SubType {
157 	short	stype;		/* [=+-?%#] action after expanded word */
158 	short	base;		/* begin position of expanded word */
159 	short	f;		/* saved value of f (DOPAT, etc) */
160 	struct tbl *var;	/* variable for ${var..} */
161 	short	quote;		/* saved value of quote (for ${..[%#]..}) */
162 	struct SubType *prev;	/* old type */
163 	struct SubType *next;	/* poped type (to avoid re-allocating) */
164 } SubType;
165 
166 void
167 expand(cp, wp, f)
168 	char *cp;		/* input word */
169 	register XPtrV *wp;	/* output words */
170 	int f;			/* DO* flags */
171 {
172 	register int UNINITIALIZED(c);
173 	register int type;	/* expansion type */
174 	register int quote = 0;	/* quoted */
175 	XString ds;		/* destination string */
176 	register char *dp, *sp;	/* dest., source */
177 	int fdo, word;		/* second pass flags; have word */
178 	int doblank;		/* field splitting of parameter/command subst */
179 	Expand x;		/* expansion variables */
180 	SubType st_head, *st;
181 	int UNINITIALIZED(newlines); /* For trailing newlines in COMSUB */
182 	int saw_eq, tilde_ok;
183 	int make_magic;
184 	size_t len;
185 
186 	if (cp == NULL)
187 		internal_errorf(1, "expand(NULL)");
188 	/* for alias, readonly, set, typeset commands */
189 	if ((f & DOVACHECK) && is_wdvarassign(cp)) {
190 		f &= ~(DOVACHECK|DOBLANK|DOGLOB|DOTILDE);
191 		f |= DOASNTILDE;
192 	}
193 	if (Flag(FNOGLOB))
194 		f &= ~DOGLOB;
195 	if (Flag(FMARKDIRS))
196 		f |= DOMARKDIRS;
197 #ifdef BRACE_EXPAND
198 	if (Flag(FBRACEEXPAND) && (f & DOGLOB))
199 		f |= DOBRACE_;
200 #endif /* BRACE_EXPAND */
201 
202 	Xinit(ds, dp, 128, ATEMP);	/* init dest. string */
203 	type = XBASE;
204 	sp = cp;
205 	fdo = 0;
206 	saw_eq = 0;
207 	tilde_ok = (f & (DOTILDE|DOASNTILDE)) ? 1 : 0; /* must be 1/0 */
208 	doblank = 0;
209 	make_magic = 0;
210 	word = (f&DOBLANK) ? IFS_WS : IFS_WORD;
211 	st_head.next = (SubType *) 0;
212 	st = &st_head;
213 
214 	while (1) {
215 		Xcheck(ds, dp);
216 
217 		switch (type) {
218 		  case XBASE:	/* original prefixed string */
219 			c = *sp++;
220 			switch (c) {
221 			  case EOS:
222 				c = 0;
223 				break;
224 			  case CHAR:
225 				c = *sp++;
226 				break;
227 			  case QCHAR:
228 				quote |= 2; /* temporary quote */
229 				c = *sp++;
230 				break;
231 			  case OQUOTE:
232 				word = IFS_WORD;
233 				tilde_ok = 0;
234 				quote = 1;
235 				continue;
236 			  case CQUOTE:
237 				quote = 0;
238 				continue;
239 			  case COMSUB:
240 				tilde_ok = 0;
241 				if (f & DONTRUNCOMMAND) {
242 					word = IFS_WORD;
243 					*dp++ = '$'; *dp++ = '(';
244 					while (*sp != '\0') {
245 						Xcheck(ds, dp);
246 						*dp++ = *sp++;
247 					}
248 					*dp++ = ')';
249 				} else {
250 					type = comsub(&x, sp);
251 					if (type == XCOM && (f&DOBLANK))
252 						doblank++;
253 					sp = strchr(sp, 0) + 1;
254 					newlines = 0;
255 				}
256 				continue;
257 			  case EXPRSUB:
258 				word = IFS_WORD;
259 				tilde_ok = 0;
260 				if (f & DONTRUNCOMMAND) {
261 					*dp++ = '$'; *dp++ = '('; *dp++ = '(';
262 					while (*sp != '\0') {
263 						Xcheck(ds, dp);
264 						*dp++ = *sp++;
265 					}
266 					*dp++ = ')'; *dp++ = ')';
267 				} else {
268 					struct tbl v;
269 					char *p;
270 
271 					v.flag = DEFINED|ISSET|INTEGER;
272 					v.type = 10; /* not default */
273 					v.name[0] = '\0';
274 					v_evaluate(&v, substitute(sp, 0),
275 						KSH_UNWIND_ERROR);
276 					sp = strchr(sp, 0) + 1;
277 					for (p = str_val(&v); *p; ) {
278 						Xcheck(ds, dp);
279 						*dp++ = *p++;
280 					}
281 				}
282 				continue;
283 			  case OSUBST: /* ${{#}var{:}[=+-?#%]word} */
284 			  /* format is:
285 			   *   OSUBST [{x] plain-variable-part \0
286 			   *     compiled-word-part CSUBST [}x]
287 			   * This is were all syntax checking gets done...
288 			   */
289 			  {
290 				char *varname = ++sp; /* skip the { or x (}) */
291 				int stype;
292 				int slen;
293 
294 				sp = strchr(sp, '\0') + 1; /* skip variable */
295 				type = varsub(&x, varname, sp, &stype, &slen);
296 				if (type < 0) {
297 					char endc;
298 					char *str, *end;
299 
300 					end = (char *) wdscan(sp, CSUBST);
301 					/* ({) the } or x is already skipped */
302 					endc = *end;
303 					*end = EOS;
304 					str = snptreef((char *) 0, 64, "%S",
305 							varname - 1);
306 					*end = endc;
307 					errorf("%s: bad substitution", str);
308 				}
309 				if (f&DOBLANK)
310 					doblank++;
311 				tilde_ok = 0;
312 				if (type == XBASE) {	/* expand? */
313 					if (!st->next) {
314 						SubType *newst;
315 
316 						newst = (SubType *) alloc(
317 							sizeof(SubType), ATEMP);
318 						newst->next = (SubType *) 0;
319 						newst->prev = st;
320 						st->next = newst;
321 					}
322 					st = st->next;
323 					st->stype = stype;
324 					st->base = Xsavepos(ds, dp);
325 					st->f = f;
326 					st->var = x.var;
327 					st->quote = quote;
328 					/* skip qualifier(s) */
329 					if (stype)
330 						sp += slen;
331 					switch (stype & 0x7f) {
332 					  case '#':
333 					  case '%':
334 						/* ! DOBLANK,DOBRACE_,DOTILDE */
335 						f = DOPAT | (f&DONTRUNCOMMAND)
336 						    | DOTEMP_;
337 						quote = 0;
338 						/* Prepend open pattern (so |
339 						 * in a trim will work as
340 						 * expected)
341 						 */
342 						*dp++ = MAGIC;
343 						*dp++ = '@' + 0x80;
344 						break;
345 					  case '=':
346 						/* Enabling tilde expansion
347 						 * after :'s here is
348 						 * non-standard ksh, but is
349 						 * consistent with rules for
350 						 * other assignments.  Not
351 						 * sure what POSIX thinks of
352 						 * this.
353 						 * Not doing tilde expansion
354 						 * for integer variables is a
355 						 * non-POSIX thing - makes
356 						 * sense though, since ~ is
357 						 * a arithmetic operator.
358 						 */
359 						if (!(x.var->flag & INTEGER))
360 							f |= DOASNTILDE|DOTILDE;
361 						f |= DOTEMP_;
362 						/* These will be done after the
363 						 * value has been assigned.
364 						 */
365 						f &= ~(DOBLANK|DOGLOB|DOBRACE_);
366 						tilde_ok = 1;
367 						break;
368 					  case '?':
369 						f &= ~DOBLANK;
370 						f |= DOTEMP_;
371 						/* fall through */
372 					  default:
373 						/* Enable tilde expansion */
374 						tilde_ok = 1;
375 						f |= DOTILDE;
376 					}
377 				} else
378 					/* skip word */
379 					sp = (char *) wdscan(sp, CSUBST);
380 				continue;
381 			  }
382 			  case CSUBST: /* only get here if expanding word */
383 				sp++; /* ({) skip the } or x */
384 				tilde_ok = 0;	/* in case of ${unset:-} */
385 				*dp = '\0';
386 				quote = st->quote;
387 				f = st->f;
388 				if (f&DOBLANK)
389 					doblank--;
390 				switch (st->stype&0x7f) {
391 				  case '#':
392 				  case '%':
393 					/* Append end-pattern */
394 					*dp++ = MAGIC; *dp++ = ')'; *dp = '\0';
395 					dp = Xrestpos(ds, dp, st->base);
396 					/* Must use st->var since calling
397 					 * global would break things
398 					 * like x[i+=1].
399 					 */
400 					x.str = trimsub(str_val(st->var),
401 						dp, st->stype);
402 					type = XSUB;
403 					if (f&DOBLANK)
404 						doblank++;
405 					st = st->prev;
406 					continue;
407 				  case '=':
408 					/* Restore our position and substitute
409 					 * the value of st->var (may not be
410 					 * the assigned value in the presence
411 					 * of integer/right-adj/etc attributes).
412 					 */
413 					dp = Xrestpos(ds, dp, st->base);
414 					/* Must use st->var since calling
415 					 * global would cause with things
416 					 * like x[i+=1] to be evaluated twice.
417 					 */
418 					/* Note: not exported by FEXPORT
419 					 * in at&t ksh.
420 					 */
421 					/* XXX POSIX says readonly is only
422 					 * fatal for special builtins (setstr
423 					 * does readonly check).
424 					 */
425 					len = strlen(dp) + 1;
426 					setstr(st->var,
427 					    debunk((char *) alloc(len, ATEMP),
428 						dp, len),
429 					    KSH_UNWIND_ERROR);
430 					x.str = str_val(st->var);
431 					type = XSUB;
432 					if (f&DOBLANK)
433 						doblank++;
434 					st = st->prev;
435 					continue;
436 				  case '?':
437 				    {
438 					char *s = Xrestpos(ds, dp, st->base);
439 
440 					errorf("%s: %s", st->var->name,
441 					    dp == s ?
442 					      "parameter null or not set"
443 					    : (debunk(s, s, strlen(s) + 1), s));
444 				    }
445 				}
446 				st = st->prev;
447 				type = XBASE;
448 				continue;
449 
450 			  case OPAT: /* open pattern: *(foo|bar) */
451 				/* Next char is the type of pattern */
452 				make_magic = 1;
453 				c = *sp++ + 0x80;
454 				break;
455 
456 			  case SPAT: /* pattern separator (|) */
457 				make_magic = 1;
458 				c = '|';
459 				break;
460 
461 			  case CPAT: /* close pattern */
462 				make_magic = 1;
463 				c = /*(*/ ')';
464 				break;
465 			}
466 			break;
467 
468 		  case XNULLSUB:
469 			/* Special case for "$@" (and "${foo[@]}") - no
470 			 * word is generated if $# is 0 (unless there is
471 			 * other stuff inside the quotes).
472 			 */
473 			type = XBASE;
474 			if (f&DOBLANK) {
475 				doblank--;
476 				/* not really correct: x=; "$x$@" should
477 				 * generate a null argument and
478 				 * set A; "${@:+}" shouldn't.
479 				 */
480 				if (dp == Xstring(ds, dp))
481 					word = IFS_WS;
482 			}
483 			continue;
484 
485 		  case XSUB:
486 			if ((c = *x.str++) == 0) {
487 				type = XBASE;
488 				if (f&DOBLANK)
489 					doblank--;
490 				continue;
491 			}
492 			break;
493 
494 		  case XARGSEP:
495 			type = XARG;
496 			quote = 1;
497 		  case XARG:
498 			if ((c = *x.str++) == '\0') {
499 				/* force null words to be created so
500 				 * set -- '' 2 ''; foo "$@" will do
501 				 * the right thing
502 				 */
503 				if (quote && x.split)
504 					word = IFS_WORD;
505 				if ((x.str = *x.u.strv++) == NULL) {
506 					type = XBASE;
507 					if (f&DOBLANK)
508 						doblank--;
509 					continue;
510 				}
511 				c = ifs0;
512 				if (c == 0) {
513 					if (quote && !x.split)
514 						continue;
515 					c = ' ';
516 				}
517 				if (quote && x.split) {
518 					/* terminate word for "$@" */
519 					type = XARGSEP;
520 					quote = 0;
521 				}
522 			}
523 			break;
524 
525 		  case XCOM:
526 			if (newlines) {		/* Spit out saved nl's */
527 				c = '\n';
528 				--newlines;
529 			} else {
530 				while ((c = shf_getc(x.u.shf)) == 0 || c == '\n')
531 				    if (c == '\n')
532 					    newlines++;	/* Save newlines */
533 				if (newlines && c != EOF) {
534 					shf_ungetc(c, x.u.shf);
535 					c = '\n';
536 					--newlines;
537 				}
538 			}
539 			if (c == EOF) {
540 				newlines = 0;
541 				shf_close(x.u.shf);
542 				if (x.split)
543 					subst_exstat = waitlast();
544 				type = XBASE;
545 				if (f&DOBLANK)
546 					doblank--;
547 				continue;
548 			}
549 			break;
550 		}
551 
552 		/* check for end of word or IFS separation */
553 		if (c == 0 || (!quote && (f & DOBLANK) && doblank && !make_magic
554 			       && ctype(c, C_IFS)))
555 		{
556 			/* How words are broken up:
557 			 *		   |       value of c
558 			 *	  word	   |	ws	nws	0
559 			 *	-----------------------------------
560 			 *	IFS_WORD	w/WS	w/NWS	w
561 			 *	IFS_WS		-/WS	w/NWS	-
562 			 *	IFS_NWS		-/NWS	w/NWS	w
563 			 *   (w means generate a word)
564 			 * Note that IFS_NWS/0 generates a word (at&t ksh
565 			 * doesn't do this, but POSIX does).
566 			 */
567 			if (word == IFS_WORD
568 			    || (!ctype(c, C_IFSWS) && (c || word == IFS_NWS)))
569 			{
570 				char *p;
571 
572 				*dp++ = '\0';
573 				p = Xclose(ds, dp);
574 #ifdef BRACE_EXPAND
575 				if (fdo & DOBRACE_)
576 					/* also does globbing */
577 					alt_expand(wp, p, p,
578 						   p + Xlength(ds, (dp - 1)),
579 						   fdo | (f & DOMARKDIRS));
580 				else
581 #endif /* BRACE_EXPAND */
582 				if (fdo & DOGLOB)
583 					glob(p, wp, f & DOMARKDIRS);
584 				else if ((f & DOPAT) || !(fdo & DOMAGIC_))
585 					XPput(*wp, p);
586 				else
587 					XPput(*wp, debunk(p, p, strlen(p) + 1));
588 				fdo = 0;
589 				saw_eq = 0;
590 				tilde_ok = (f & (DOTILDE|DOASNTILDE)) ? 1 : 0;
591 				if (c != 0)
592 					Xinit(ds, dp, 128, ATEMP);
593 			}
594 			if (c == 0)
595 				return;
596 			if (word != IFS_NWS)
597 				word = ctype(c, C_IFSWS) ? IFS_WS : IFS_NWS;
598 		} else {
599 			/* age tilde_ok info - ~ code tests second bit */
600 			tilde_ok <<= 1;
601 			/* mark any special second pass chars */
602 			if (!quote)
603 				switch (c) {
604 				  case '[':
605 				  case NOT:
606 				  case '-':
607 				  case ']':
608 					/* For character classes - doesn't hurt
609 					 * to have magic !,-,]'s outside of
610 					 * [...] expressions.
611 					 */
612 					if (f & (DOPAT | DOGLOB)) {
613 						fdo |= DOMAGIC_;
614 						if (c == '[')
615 							fdo |= f & DOGLOB;
616 						*dp++ = MAGIC;
617 					}
618 					break;
619 				  case '*':
620 				  case '?':
621 					if (f & (DOPAT | DOGLOB)) {
622 						fdo |= DOMAGIC_ | (f & DOGLOB);
623 						*dp++ = MAGIC;
624 					}
625 					break;
626 #ifdef BRACE_EXPAND
627 				  case OBRACE:
628 				  case ',':
629 				  case CBRACE:
630 					if ((f & DOBRACE_) && (c == OBRACE
631 						|| (fdo & DOBRACE_)))
632 					{
633 						fdo |= DOBRACE_|DOMAGIC_;
634 						*dp++ = MAGIC;
635 					}
636 					break;
637 #endif /* BRACE_EXPAND */
638 				  case '=':
639 					/* Note first unquoted = for ~ */
640 					if (!(f & DOTEMP_) && !saw_eq) {
641 						saw_eq = 1;
642 						tilde_ok = 1;
643 					}
644 					break;
645 				  case PATHSEP: /* : */
646 					/* Note unquoted : for ~ */
647 					if (!(f & DOTEMP_) && (f & DOASNTILDE))
648 						tilde_ok = 1;
649 					break;
650 				  case '~':
651 					/* tilde_ok is reset whenever
652 					 * any of ' " $( $(( ${ } are seen.
653 					 * Note that tilde_ok must be preserved
654 					 * through the sequence ${A=a=}~
655 					 */
656 					if (type == XBASE
657 					    && (f & (DOTILDE|DOASNTILDE))
658 					    && (tilde_ok & 2))
659 					{
660 						char *p, *dp_x;
661 
662 						dp_x = dp;
663 						p = maybe_expand_tilde(sp,
664 							&ds, &dp_x,
665 							f & DOASNTILDE);
666 						if (p) {
667 							if (dp != dp_x)
668 								word = IFS_WORD;
669 							dp = dp_x;
670 							sp = p;
671 							continue;
672 						}
673 					}
674 					break;
675 				}
676 			else
677 				quote &= ~2; /* undo temporary */
678 
679 			if (make_magic) {
680 				make_magic = 0;
681 				fdo |= DOMAGIC_ | (f & DOGLOB);
682 				*dp++ = MAGIC;
683 			} else if (ISMAGIC(c)) {
684 				fdo |= DOMAGIC_;
685 				*dp++ = MAGIC;
686 			}
687 			*dp++ = c; /* save output char */
688 			word = IFS_WORD;
689 		}
690 	}
691 }
692 
693 /*
694  * Prepare to generate the string returned by ${} substitution.
695  */
696 static int
697 varsub(xp, sp, word, stypep, slenp)
698 	Expand *xp;
699 	char *sp;
700 	char *word;
701 	int *stypep;	/* becomes qualifier type */
702 	int *slenp;	/* " " len (=, :=, etc.) valid iff *stypep != 0 */
703 {
704 	int c;
705 	int state;	/* next state: XBASE, XARG, XSUB, XNULLSUB */
706 	int stype;	/* substitution type */
707 	int slen;
708 	char *p;
709 	struct tbl *vp;
710 
711 	if (sp[0] == '\0')	/* Bad variable name */
712 		return -1;
713 
714 	xp->var = (struct tbl *) 0;
715 
716 	/* ${#var}, string length or array size */
717 	if (sp[0] == '#' && (c = sp[1]) != '\0') {
718 		int zero_ok = 0;
719 
720 		/* Can't have any modifiers for ${#...} */
721 		if (*word != CSUBST)
722 			return -1;
723 		sp++;
724 		/* Check for size of array */
725 		if ((p=strchr(sp,'[')) && (p[1]=='*'||p[1]=='@') && p[2]==']') {
726 			int n = 0;
727 			int max = 0;
728 			vp = global(arrayname(sp));
729 			if (vp->flag & (ISSET|ARRAY))
730 				zero_ok = 1;
731 			for (; vp; vp = vp->u.array)
732 				if (vp->flag & ISSET) {
733 					max = vp->index + 1;
734 					n++;
735 				}
736 			c = n; /* ksh88/ksh93 go for number, not max index */
737 		} else if (c == '*' || c == '@')
738 			c = e->loc->argc;
739 		else {
740 			p = str_val(global(sp));
741 			zero_ok = p != null;
742 			c = strlen(p);
743 		}
744 		if (Flag(FNOUNSET) && c == 0 && !zero_ok)
745 			errorf("%s: parameter not set", sp);
746 		*stypep = 0; /* unqualified variable/string substitution */
747 		xp->str = str_save(ulton((unsigned long)c, 10), ATEMP);
748 		return XSUB;
749 	}
750 
751 	/* Check for qualifiers in word part */
752 	stype = 0;
753 	c = word[slen = 0] == CHAR ? word[1] : 0;
754 	if (c == ':') {
755 		slen += 2;
756 		stype = 0x80;
757 		c = word[slen + 0] == CHAR ? word[slen + 1] : 0;
758 	}
759 	if (ctype(c, C_SUBOP1)) {
760 		slen += 2;
761 		stype |= c;
762 	} else if (ctype(c, C_SUBOP2)) { /* Note: ksh88 allows :%, :%%, etc */
763 		slen += 2;
764 		stype = c;
765 		if (word[slen + 0] == CHAR && c == word[slen + 1]) {
766 			stype |= 0x80;
767 			slen += 2;
768 		}
769 	} else if (stype)	/* : is not ok */
770 		return -1;
771 	if (!stype && *word != CSUBST)
772 		return -1;
773 	*stypep = stype;
774 	*slenp = slen;
775 
776 	c = sp[0];
777 	if (c == '*' || c == '@') {
778 		switch (stype & 0x7f) {
779 		  case '=':	/* can't assign to a vector */
780 		  case '%':	/* can't trim a vector (yet) */
781 		  case '#':
782 			return -1;
783 		}
784 		if (e->loc->argc == 0) {
785 			xp->str = null;
786 			state = c == '@' ? XNULLSUB : XSUB;
787 		} else {
788 			xp->u.strv = (const char **) e->loc->argv + 1;
789 			xp->str = *xp->u.strv++;
790 			xp->split = c == '@'; /* $@ */
791 			state = XARG;
792 		}
793 	} else {
794 		if ((p=strchr(sp,'[')) && (p[1]=='*'||p[1]=='@') && p[2]==']') {
795 			XPtrV wv;
796 
797 			switch (stype & 0x7f) {
798 			  case '=':	/* can't assign to a vector */
799 			  case '%':	/* can't trim a vector (yet) */
800 			  case '#':
801 				return -1;
802 			}
803 			XPinit(wv, 32);
804 			vp = global(arrayname(sp));
805 			for (; vp; vp = vp->u.array) {
806 				if (!(vp->flag&ISSET))
807 					continue;
808 				XPput(wv, str_val(vp));
809 			}
810 			if (XPsize(wv) == 0) {
811 				xp->str = null;
812 				state = p[1] == '@' ? XNULLSUB : XSUB;
813 				XPfree(wv);
814 			} else {
815 				XPput(wv, 0);
816 				xp->u.strv = (const char **) XPptrv(wv);
817 				xp->str = *xp->u.strv++;
818 				xp->split = p[1] == '@'; /* ${foo[@]} */
819 				state = XARG;
820 			}
821 		} else {
822 			/* Can't assign things like $! or $1 */
823 			if ((stype & 0x7f) == '='
824 			    && (ctype(*sp, C_VAR1) || digit(*sp)))
825 				return -1;
826 			xp->var = global(sp);
827 			xp->str = str_val(xp->var);
828 			state = XSUB;
829 		}
830 	}
831 
832 	c = stype&0x7f;
833 	/* test the compiler's code generator */
834 	if (ctype(c, C_SUBOP2) ||
835 	    (((stype&0x80) ? *xp->str=='\0' : xp->str==null) ? /* undef? */
836 	     c == '=' || c == '-' || c == '?' : c == '+'))
837 		state = XBASE;	/* expand word instead of variable value */
838 	if (Flag(FNOUNSET) && xp->str == null
839 	    && (ctype(c, C_SUBOP2) || (state != XBASE && c != '+')))
840 		errorf("%s: parameter not set", sp);
841 	return state;
842 }
843 
844 /*
845  * Run the command in $(...) and read its output.
846  */
847 static int
848 comsub(xp, cp)
849 	register Expand *xp;
850 	char *cp;
851 {
852 	Source *s, *sold;
853 	register struct op *t;
854 	struct shf *shf;
855 
856 	s = pushs(SSTRING, ATEMP);
857 	s->start = s->str = cp;
858 	sold = source;
859 	t = compile(s);
860 	source = sold;
861 
862 	if (t == NULL)
863 		return XBASE;
864 
865 	if (t != NULL && t->type == TCOM && /* $(<file) */
866 	    *t->args == NULL && *t->vars == NULL && t->ioact != NULL) {
867 		register struct ioword *io = *t->ioact;
868 		char *name;
869 
870 		if ((io->flag&IOTYPE) != IOREAD)
871 			errorf("funny $() command: %s",
872 				snptreef((char *) 0, 32, "%R", io));
873 		shf = shf_open(name = evalstr(io->name, DOTILDE), O_RDONLY, 0,
874 			SHF_MAPHI|SHF_CLEXEC);
875 		if (shf == NULL)
876 			errorf("%s: cannot open $() input", name);
877 		xp->split = 0;	/* no waitlast() */
878 	} else {
879 		int ofd1, pv[2];
880 		openpipe(pv);
881 		shf = shf_fdopen(pv[0], SHF_RD, (struct shf *) 0);
882 		ofd1 = savefd(1, 0);	/* fd 1 may be closed... */
883 		if (pv[1] != 1) {
884 			ksh_dup2(pv[1], 1, FALSE);
885 			close(pv[1]);
886 		}
887 		execute(t, XFORK|XXCOM|XPIPEO);
888 		restfd(1, ofd1);
889 		startlast();
890 		xp->split = 1;	/* waitlast() */
891 	}
892 
893 	xp->u.shf = shf;
894 	return XCOM;
895 }
896 
897 /*
898  * perform #pattern and %pattern substitution in ${}
899  */
900 
901 static char *
902 trimsub(str, pat, how)
903 	register char *str;
904 	char *pat;
905 	int how;
906 {
907 	register char *end = strchr(str, 0);
908 	register char *p, c;
909 
910 	switch (how&0xff) {	/* UCHAR_MAX maybe? */
911 	  case '#':		/* shortest at beginning */
912 		for (p = str; p <= end; p++) {
913 			c = *p; *p = '\0';
914 			if (gmatch(str, pat, FALSE)) {
915 				*p = c;
916 				return p;
917 			}
918 			*p = c;
919 		}
920 		break;
921 	  case '#'|0x80:	/* longest match at beginning */
922 		for (p = end; p >= str; p--) {
923 			c = *p; *p = '\0';
924 			if (gmatch(str, pat, FALSE)) {
925 				*p = c;
926 				return p;
927 			}
928 			*p = c;
929 		}
930 		break;
931 	  case '%':		/* shortest match at end */
932 		for (p = end; p >= str; p--) {
933 			if (gmatch(p, pat, FALSE))
934 				return str_nsave(str, p - str, ATEMP);
935 		}
936 		break;
937 	  case '%'|0x80:	/* longest match at end */
938 		for (p = str; p <= end; p++) {
939 			if (gmatch(p, pat, FALSE))
940 				return str_nsave(str, p - str, ATEMP);
941 		}
942 		break;
943 	}
944 
945 	return str;		/* no match, return string */
946 }
947 
948 /*
949  * glob
950  * Name derived from V6's /etc/glob, the program that expanded filenames.
951  */
952 
953 /* XXX cp not const 'cause slashes are temporarily replaced with nulls... */
954 static void
955 glob(cp, wp, markdirs)
956 	char *cp;
957 	register XPtrV *wp;
958 	int markdirs;
959 {
960 	int oldsize = XPsize(*wp);
961 
962 	if (glob_str(cp, wp, markdirs) == 0)
963 		XPput(*wp, debunk(cp, cp, strlen(cp) + 1));
964 	else
965 		qsortp(XPptrv(*wp) + oldsize, (size_t)(XPsize(*wp) - oldsize),
966 			xstrcmp);
967 }
968 
969 #define GF_NONE		0
970 #define GF_EXCHECK	BIT(0)		/* do existence check on file */
971 #define GF_GLOBBED	BIT(1)		/* some globbing has been done */
972 #define GF_MARKDIR	BIT(2)		/* add trailing / to directories */
973 
974 /* Apply file globbing to cp and store the matching files in wp.  Returns
975  * the number of matches found.
976  */
977 int
978 glob_str(cp, wp, markdirs)
979 	char *cp;
980 	XPtrV *wp;
981 	int markdirs;
982 {
983 	int oldsize = XPsize(*wp);
984 	XString xs;
985 	char *xp;
986 
987 	Xinit(xs, xp, 256, ATEMP);
988 	globit(&xs, &xp, cp, wp, markdirs ? GF_MARKDIR : GF_NONE);
989 	Xfree(xs, xp);
990 
991 	return XPsize(*wp) - oldsize;
992 }
993 
994 static void
995 globit(xs, xpp, sp, wp, check)
996 	XString *xs;		/* dest string */
997 	char **xpp;		/* ptr to dest end */
998 	char *sp;		/* source path */
999 	register XPtrV *wp;	/* output list */
1000 	int check;		/* GF_* flags */
1001 {
1002 	register char *np;	/* next source component */
1003 	char *xp = *xpp;
1004 	char *se;
1005 	char odirsep;
1006 
1007 	/* This to allow long expansions to be interrupted */
1008 	intrcheck();
1009 
1010 	if (sp == NULL) {	/* end of source path */
1011 		/* We only need to check if the file exists if a pattern
1012 		 * is followed by a non-pattern (eg, foo*x/bar; no check
1013 		 * is needed for foo* since the match must exist) or if
1014 		 * any patterns were expanded and the markdirs option is set.
1015 		 * Symlinks make things a bit tricky...
1016 		 */
1017 		if ((check & GF_EXCHECK)
1018 		    || ((check & GF_MARKDIR) && (check & GF_GLOBBED)))
1019 		{
1020 #define stat_check()	(stat_done ? stat_done : \
1021 			    (stat_done = stat(Xstring(*xs, xp), &statb) < 0 \
1022 				? -1 : 1))
1023 			struct stat lstatb, statb;
1024 			int stat_done = 0;	 /* -1: failed, 1 ok */
1025 
1026 			if (lstat(Xstring(*xs, xp), &lstatb) < 0)
1027 				return;
1028 			/* special case for systems which strip trailing
1029 			 * slashes from regular files (eg, /etc/passwd/).
1030 			 * SunOS 4.1.3 does this...
1031 			 */
1032 			if ((check & GF_EXCHECK) && xp > Xstring(*xs, xp)
1033 			    && ISDIRSEP(xp[-1]) && !S_ISDIR(lstatb.st_mode)
1034 #ifdef S_ISLNK
1035 			    && (!S_ISLNK(lstatb.st_mode)
1036 				|| stat_check() < 0
1037 				|| !S_ISDIR(statb.st_mode))
1038 #endif /* S_ISLNK */
1039 				)
1040 				return;
1041 			/* Possibly tack on a trailing / if there isn't already
1042 			 * one and if the file is a directory or a symlink to a
1043 			 * directory
1044 			 */
1045 			if (((check & GF_MARKDIR) && (check & GF_GLOBBED))
1046 			    && xp > Xstring(*xs, xp) && !ISDIRSEP(xp[-1])
1047 			    && (S_ISDIR(lstatb.st_mode)
1048 #ifdef S_ISLNK
1049 				|| (S_ISLNK(lstatb.st_mode)
1050 				    && stat_check() > 0
1051 				    && S_ISDIR(statb.st_mode))
1052 #endif /* S_ISLNK */
1053 				    ))
1054 			{
1055 				*xp++ = DIRSEP;
1056 				*xp = '\0';
1057 			}
1058 		}
1059 #ifdef OS2 /* Done this way to avoid bug in gcc 2.7.2... */
1060     /* Ugly kludge required for command
1061      * completion - see how search_access()
1062      * is implemented for OS/2...
1063      */
1064 # define KLUDGE_VAL	4
1065 #else /* OS2 */
1066 # define KLUDGE_VAL	0
1067 #endif /* OS2 */
1068 		XPput(*wp, str_nsave(Xstring(*xs, xp), Xlength(*xs, xp)
1069 			+ KLUDGE_VAL, ATEMP));
1070 		return;
1071 	}
1072 
1073 	if (xp > Xstring(*xs, xp))
1074 		*xp++ = DIRSEP;
1075 	while (ISDIRSEP(*sp)) {
1076 		Xcheck(*xs, xp);
1077 		*xp++ = *sp++;
1078 	}
1079 	np = ksh_strchr_dirsep(sp);
1080 	if (np != NULL) {
1081 		se = np;
1082 		odirsep = *np;	/* don't assume DIRSEP, can be multiple kinds */
1083 		*np++ = '\0';
1084 	} else {
1085 		odirsep = '\0'; /* keep gcc quiet */
1086 		se = sp + strlen(sp);
1087 	}
1088 
1089 
1090 	/* Check if sp needs globbing - done to avoid pattern checks for strings
1091 	 * containing MAGIC characters, open ['s without the matching close ],
1092 	 * etc. (otherwise opendir() will be called which may fail because the
1093 	 * directory isn't readable - if no globbing is needed, only execute
1094 	 * permission should be required (as per POSIX)).
1095 	 */
1096 	if (!has_globbing(sp, se)) {
1097 		XcheckN(*xs, xp, se - sp + 1);
1098 		debunk(xp, sp, Xnleft(*xs, xp));
1099 		xp += strlen(xp);
1100 		*xpp = xp;
1101 		globit(xs, xpp, np, wp, check);
1102 	} else {
1103 		DIR *dirp;
1104 		struct dirent *d;
1105 		char *name;
1106 		int len;
1107 		int prefix_len;
1108 
1109 		/* xp = *xpp;	   copy_non_glob() may have re-alloc'd xs */
1110 		*xp = '\0';
1111 		prefix_len = Xlength(*xs, xp);
1112 		dirp = ksh_opendir(prefix_len ? Xstring(*xs, xp) : ".");
1113 		if (dirp == NULL)
1114 			goto Nodir;
1115 		while ((d = readdir(dirp)) != NULL) {
1116 			name = d->d_name;
1117 			if (name[0] == '.' &&
1118 			    (name[1] == 0 || (name[1] == '.' && name[2] == 0)))
1119 				continue; /* always ignore . and .. */
1120 			if ((*name == '.' && *sp != '.')
1121 			    || !gmatch(name, sp, TRUE))
1122 				continue;
1123 
1124 			len = NLENGTH(d) + 1;
1125 			XcheckN(*xs, xp, len);
1126 			memcpy(xp, name, len);
1127 			*xpp = xp + len - 1;
1128 			globit(xs, xpp, np, wp,
1129 				(check & GF_MARKDIR) | GF_GLOBBED
1130 				| (np ? GF_EXCHECK : GF_NONE));
1131 			xp = Xstring(*xs, xp) + prefix_len;
1132 		}
1133 		closedir(dirp);
1134 	  Nodir:;
1135 	}
1136 
1137 	if (np != NULL)
1138 		*--np = odirsep;
1139 }
1140 
1141 #if 0
1142 /* Check if p contains something that needs globbing; if it does, 0 is
1143  * returned; if not, p is copied into xs/xp after stripping any MAGICs
1144  */
1145 static int	copy_non_glob ARGS((XString *xs, char **xpp, char *p));
1146 static int
1147 copy_non_glob(xs, xpp, p)
1148 	XString *xs;
1149 	char **xpp;
1150 	char *p;
1151 {
1152 	char *xp;
1153 	int len = strlen(p);
1154 
1155 	XcheckN(*xs, *xpp, len);
1156 	xp = *xpp;
1157 	for (; *p; p++) {
1158 		if (ISMAGIC(*p)) {
1159 			int c = *++p;
1160 
1161 			if (c == '*' || c == '?')
1162 				return 0;
1163 			if (*p == '[') {
1164 				char *q = p + 1;
1165 
1166 				if (ISMAGIC(*q) && q[1] == NOT)
1167 					q += 2;
1168 				if (ISMAGIC(*q) && q[1] == ']')
1169 					q += 2;
1170 				for (; *q; q++)
1171 					if (ISMAGIC(*q) && *++q == ']')
1172 						return 0;
1173 				/* pass a literal [ through */
1174 			}
1175 			/* must be a MAGIC-MAGIC, or MAGIC-!, MAGIC--, etc. */
1176 		}
1177 		*xp++ = *p;
1178 	}
1179 	*xp = '\0';
1180 	*xpp = xp;
1181 	return 1;
1182 }
1183 #endif /* 0 */
1184 
1185 /* remove MAGIC from string */
1186 char *
1187 debunk(dp, sp, dlen)
1188 	char *dp;
1189 	const char *sp;
1190 	size_t dlen;
1191 {
1192 	char *d, *s;
1193 
1194 	if ((s = strchr(sp, MAGIC))) {
1195 		if (s - sp >= dlen)
1196 			return dp;
1197 		memcpy(dp, sp, s - sp);
1198 		for (d = dp + (s - sp); *s && (d - dp < dlen); s++)
1199 			if (!ISMAGIC(*s) || !(*++s & 0x80)
1200 			    || !strchr("*+?@! ", *s & 0x7f))
1201 				*d++ = *s;
1202 			else {
1203 				/* extended pattern operators: *+?@! */
1204 				if ((*s & 0x7f) != ' ')
1205 					*d++ = *s & 0x7f;
1206 				if (d - dp < dlen)
1207 					*d++ = '(';
1208 			}
1209 		*d = '\0';
1210 	} else if (dp != sp)
1211 		strlcpy(dp, sp, dlen);
1212 	return dp;
1213 }
1214 
1215 /* Check if p is an unquoted name, possibly followed by a / or :.  If so
1216  * puts the expanded version in *dcp,dp and returns a pointer in p just
1217  * past the name, otherwise returns 0.
1218  */
1219 static char *
1220 maybe_expand_tilde(p, dsp, dpp, isassign)
1221 	char *p;
1222 	XString *dsp;
1223 	char **dpp;
1224 	int isassign;
1225 {
1226 	XString ts;
1227 	char *dp = *dpp;
1228 	char *tp, *r;
1229 
1230 	Xinit(ts, tp, 16, ATEMP);
1231 	/* : only for DOASNTILDE form */
1232 	while (p[0] == CHAR && !ISDIRSEP(p[1])
1233 	       && (!isassign || p[1] != PATHSEP))
1234 	{
1235 		Xcheck(ts, tp);
1236 		*tp++ = p[1];
1237 		p += 2;
1238 	}
1239 	*tp = '\0';
1240 	r = (p[0] == EOS || p[0] == CHAR || p[0] == CSUBST) ? tilde(Xstring(ts, tp)) : (char *) 0;
1241 	Xfree(ts, tp);
1242 	if (r) {
1243 		while (*r) {
1244 			Xcheck(*dsp, dp);
1245 			if (ISMAGIC(*r))
1246 				*dp++ = MAGIC;
1247 			*dp++ = *r++;
1248 		}
1249 		*dpp = dp;
1250 		r = p;
1251 	}
1252 	return r;
1253 }
1254 
1255 /*
1256  * tilde expansion
1257  *
1258  * based on a version by Arnold Robbins
1259  */
1260 
1261 static char *
1262 tilde(cp)
1263 	char *cp;
1264 {
1265 	char *dp;
1266 
1267 	if (cp[0] == '\0')
1268 		dp = str_val(global("HOME"));
1269 	else if (cp[0] == '+' && cp[1] == '\0')
1270 		dp = str_val(global("PWD"));
1271 	else if (cp[0] == '-' && cp[1] == '\0')
1272 		dp = str_val(global("OLDPWD"));
1273 	else
1274 		dp = homedir(cp);
1275 	/* If HOME, PWD or OLDPWD are not set, don't expand ~ */
1276 	if (dp == null)
1277 		dp = (char *) 0;
1278 	return dp;
1279 }
1280 
1281 /*
1282  * map userid to user's home directory.
1283  * note that 4.3's getpw adds more than 6K to the shell,
1284  * and the YP version probably adds much more.
1285  * we might consider our own version of getpwnam() to keep the size down.
1286  */
1287 
1288 static char *
1289 homedir(name)
1290 	char *name;
1291 {
1292 	register struct tbl *ap;
1293 
1294 	ap = tenter(&homedirs, name, hash(name));
1295 	if (!(ap->flag & ISSET)) {
1296 #ifdef OS2
1297 		/* No usernames in OS2 - punt */
1298 		return NULL;
1299 #else /* OS2 */
1300 		struct passwd *pw;
1301 
1302 		pw = getpwnam(name);
1303 		if (pw == NULL)
1304 			return NULL;
1305 		ap->val.s = str_save(pw->pw_dir, APERM);
1306 		ap->flag |= DEFINED|ISSET|ALLOC;
1307 #endif /* OS2 */
1308 	}
1309 	return ap->val.s;
1310 }
1311 
1312 #ifdef BRACE_EXPAND
1313 static void
1314 alt_expand(wp, start, exp_start, end, fdo)
1315 	XPtrV *wp;
1316 	char *start, *exp_start;
1317 	char *end;
1318 	int fdo;
1319 {
1320 	int UNINITIALIZED(count);
1321 	char *brace_start, *brace_end, *UNINITIALIZED(comma);
1322 	char *field_start;
1323 	char *p;
1324 
1325 	/* search for open brace */
1326 	for (p = exp_start; (p = strchr(p, MAGIC)) && p[1] != OBRACE; p += 2)
1327 		;
1328 	brace_start = p;
1329 
1330 	/* find matching close brace, if any */
1331 	if (p) {
1332 		comma = (char *) 0;
1333 		count = 1;
1334 		for (p += 2; *p && count; p++) {
1335 			if (ISMAGIC(*p)) {
1336 				if (*++p == OBRACE)
1337 					count++;
1338 				else if (*p == CBRACE)
1339 					--count;
1340 				else if (*p == ',' && count == 1)
1341 					comma = p;
1342 			}
1343 		}
1344 	}
1345 	/* no valid expansions... */
1346 	if (!p || count != 0) {
1347 		/* Note that given a{{b,c} we do not expand anything (this is
1348 		 * what at&t ksh does.  This may be changed to do the {b,c}
1349 		 * expansion. }
1350 		 */
1351 		if (fdo & DOGLOB)
1352 			glob(start, wp, fdo & DOMARKDIRS);
1353 		else
1354 			XPput(*wp, debunk(start, start, end - start));
1355 		return;
1356 	}
1357 	brace_end = p;
1358 	if (!comma) {
1359 		alt_expand(wp, start, brace_end, end, fdo);
1360 		return;
1361 	}
1362 
1363 	/* expand expression */
1364 	field_start = brace_start + 2;
1365 	count = 1;
1366 	for (p = brace_start + 2; p != brace_end; p++) {
1367 		if (ISMAGIC(*p)) {
1368 			if (*++p == OBRACE)
1369 				count++;
1370 			else if ((*p == CBRACE && --count == 0)
1371 				 || (*p == ',' && count == 1))
1372 			{
1373 				char *new;
1374 				int l1, l2, l3;
1375 
1376 				l1 = brace_start - start;
1377 				l2 = (p - 1) - field_start;
1378 				l3 = end - brace_end;
1379 				new = (char *) alloc(l1 + l2 + l3 + 1, ATEMP);
1380 				memcpy(new, start, l1);
1381 				memcpy(new + l1, field_start, l2);
1382 				memcpy(new + l1 + l2, brace_end, l3);
1383 				new[l1 + l2 + l3] = '\0';
1384 				alt_expand(wp, new, new + l1,
1385 					   new + l1 + l2 + l3, fdo);
1386 				field_start = p + 1;
1387 			}
1388 		}
1389 	}
1390 	return;
1391 }
1392 #endif /* BRACE_EXPAND */
1393