xref: /plan9/sys/src/ape/cmd/pdksh/eval.c (revision 7dd7cddf99dd7472612f1413b4da293630e6b1bc)
1 /*
2  * Expansion - quoting, separation, substitution, globbing
3  */
4 
5 #include "sh.h"
6 #include <pwd.h>
7 #include "ksh_dir.h"
8 #include "ksh_stat.h"
9 
10 /*
11  * string expansion
12  *
13  * first pass: quoting, IFS separation, ~, ${}, $() and $(()) substitution.
14  * second pass: alternation ({,}), filename expansion (*?[]).
15  */
16 
17 /* expansion generator state */
18 typedef struct Expand {
19 	/* int  type; */	/* see expand() */
20 	const char *str;	/* string */
21 	union {
22 		const char **strv;/* string[] */
23 		struct shf *shf;/* file */
24 	} u;			/* source */
25 	struct tbl *var;	/* variable in ${var..} */
26 	short	split;		/* split "$@" / call waitlast $() */
27 } Expand;
28 
29 #define	XBASE		0	/* scanning original */
30 #define	XSUB		1	/* expanding ${} string */
31 #define	XARGSEP		2	/* ifs0 between "$*" */
32 #define	XARG		3	/* expanding $*, $@ */
33 #define	XCOM		4	/* expanding $() */
34 #define XNULLSUB	5	/* "$@" when $# is 0 (don't generate word) */
35 
36 /* States used for field splitting */
37 #define IFS_WORD	0	/* word has chars (or quotes) */
38 #define IFS_WS		1	/* have seen IFS white-space */
39 #define IFS_NWS		2	/* have seen IFS non-white-space */
40 
41 static	int	varsub ARGS((Expand *xp, char *sp, char *word, int *stypep, int *slenp));
42 static	int	comsub ARGS((Expand *xp, char *cp));
43 static	char   *trimsub ARGS((char *str, char *pat, int how));
44 static	void	glob ARGS((char *cp, XPtrV *wp, int markdirs));
45 static	void	globit ARGS((XString *xs, char **xpp, char *sp, XPtrV *wp,
46 			     int check));
47 static char	*maybe_expand_tilde ARGS((char *p, XString *dsp, char **dpp,
48 					  int isassign));
49 static	char   *tilde ARGS((char *acp));
50 static	char   *homedir ARGS((char *name));
51 #ifdef BRACE_EXPAND
52 static void	alt_expand ARGS((XPtrV *wp, char *start, char *exp_start,
53 				 char *end, int fdo));
54 #endif
55 
56 /* compile and expand word */
57 char *
substitute(cp,f)58 substitute(cp, f)
59 	const char *cp;
60 	int f;
61 {
62 	struct source *s, *sold;
63 
64 	sold = source;
65 	s = pushs(SWSTR, ATEMP);
66 	s->start = s->str = cp;
67 	source = s;
68 	if (yylex(ONEWORD) != LWORD)
69 		internal_errorf(1, "substitute");
70 	source = sold;
71 	afree(s, ATEMP);
72 	return evalstr(yylval.cp, f);
73 }
74 
75 /*
76  * expand arg-list
77  */
78 char **
eval(ap,f)79 eval(ap, f)
80 	register char **ap;
81 	int f;
82 {
83 	XPtrV w;
84 
85 	if (*ap == NULL)
86 		return ap;
87 	XPinit(w, 32);
88 	XPput(w, NULL);		/* space for shell name */
89 #ifdef	SHARPBANG
90 	XPput(w, NULL);		/* and space for one arg */
91 #endif
92 	while (*ap != NULL)
93 		expand(*ap++, &w, f);
94 	XPput(w, NULL);
95 #ifdef	SHARPBANG
96 	return (char **) XPclose(w) + 2;
97 #else
98 	return (char **) XPclose(w) + 1;
99 #endif
100 }
101 
102 /*
103  * expand string
104  */
105 char *
evalstr(cp,f)106 evalstr(cp, f)
107 	char *cp;
108 	int f;
109 {
110 	XPtrV w;
111 
112 	XPinit(w, 1);
113 	expand(cp, &w, f);
114 	cp = (XPsize(w) == 0) ? null : (char*) *XPptrv(w);
115 	XPfree(w);
116 	return cp;
117 }
118 
119 /*
120  * expand string - return only one component
121  * used from iosetup to expand redirection files
122  */
123 char *
evalonestr(cp,f)124 evalonestr(cp, f)
125 	register char *cp;
126 	int f;
127 {
128 	XPtrV w;
129 
130 	XPinit(w, 1);
131 	expand(cp, &w, f);
132 	switch (XPsize(w)) {
133 	case 0:
134 		cp = null;
135 		break;
136 	case 1:
137 		cp = (char*) *XPptrv(w);
138 		break;
139 	default:
140 		cp = evalstr(cp, f&~DOGLOB);
141 		break;
142 	}
143 	XPfree(w);
144 	return cp;
145 }
146 
147 /* for nested substitution: ${var:=$var2} */
148 typedef struct SubType {
149 	short	stype;		/* [=+-?%#] action after expanded word */
150 	short	base;		/* begin position of expanded word */
151 	short	f;		/* saved value of f (DOPAT, etc) */
152 	struct tbl *var;	/* variable for ${var..} */
153 	short	quote;		/* saved value of quote (for ${..[%#]..}) */
154 	struct SubType *prev;	/* old type */
155 	struct SubType *next;	/* poped type (to avoid re-allocating) */
156 } SubType;
157 
158 void
expand(cp,wp,f)159 expand(cp, wp, f)
160 	char *cp;		/* input word */
161 	register XPtrV *wp;	/* output words */
162 	int f;			/* DO* flags */
163 {
164 	register int UNINITIALIZED(c);
165 	register int type;	/* expansion type */
166 	register int quote = 0;	/* quoted */
167 	XString ds;		/* destination string */
168 	register char *dp, *sp;	/* dest., source */
169 	int fdo, word;		/* second pass flags; have word */
170 	int doblank;		/* field spliting of parameter/command subst */
171 	Expand x;		/* expansion variables */
172 	SubType st_head, *st;
173 	int UNINITIALIZED(newlines); /* For trailing newlines in COMSUB */
174 	int saw_eq, tilde_ok;
175 	int make_magic;
176 
177 	if (cp == NULL)
178 		internal_errorf(1, "expand(NULL)");
179 	/* for alias, readonly, set, typeset commands */
180 	if ((f & DOVACHECK) && is_wdvarassign(cp)) {
181 		f &= ~(DOVACHECK|DOBLANK|DOGLOB|DOTILDE);
182 		f |= DOASNTILDE;
183 	}
184 	if (Flag(FNOGLOB))
185 		f &= ~DOGLOB;
186 	if (Flag(FMARKDIRS))
187 		f |= DOMARKDIRS;
188 #ifdef BRACE_EXPAND
189 	if (Flag(FBRACEEXPAND) && (f & DOGLOB))
190 		f |= DOBRACE_;
191 #endif /* BRACE_EXPAND */
192 
193 	Xinit(ds, dp, 128, ATEMP);	/* init dest. string */
194 	type = XBASE;
195 	sp = cp;
196 	fdo = 0;
197 	saw_eq = 0;
198 	tilde_ok = (f & (DOTILDE|DOASNTILDE)) ? 1 : 0; /* must be 1/0 */
199 	doblank = 0;
200 	make_magic = 0;
201 	word = (f&DOBLANK) ? IFS_WS : IFS_WORD;
202 	st_head.next = (SubType *) 0;
203 	st = &st_head;
204 
205 	while (1) {
206 		Xcheck(ds, dp);
207 
208 		switch (type) {
209 		  case XBASE:	/* original prefixed string */
210 			c = *sp++;
211 			switch (c) {
212 			  case EOS:
213 				c = 0;
214 				break;
215 			  case CHAR:
216 				c = *sp++;
217 				break;
218 			  case QCHAR:
219 				quote |= 2; /* temporary quote */
220 				c = *sp++;
221 				break;
222 			  case OQUOTE:
223 				word = IFS_WORD;
224 				tilde_ok = 0;
225 				quote = 1;
226 				continue;
227 			  case CQUOTE:
228 				quote = 0;
229 				continue;
230 			  case COMSUB:
231 				tilde_ok = 0;
232 				if (f & DONTRUNCOMMAND) {
233 					word = IFS_WORD;
234 					*dp++ = '$'; *dp++ = '(';
235 					while (*sp != '\0') {
236 						Xcheck(ds, dp);
237 						*dp++ = *sp++;
238 					}
239 					*dp++ = ')';
240 				} else {
241 					type = comsub(&x, sp);
242 					if (type == XCOM && (f&DOBLANK))
243 						doblank++;
244 					sp = strchr(sp, 0) + 1;
245 					newlines = 0;
246 				}
247 				continue;
248 			  case EXPRSUB:
249 				word = IFS_WORD;
250 				tilde_ok = 0;
251 				if (f & DONTRUNCOMMAND) {
252 					*dp++ = '$'; *dp++ = '('; *dp++ = '(';
253 					while (*sp != '\0') {
254 						Xcheck(ds, dp);
255 						*dp++ = *sp++;
256 					}
257 					*dp++ = ')'; *dp++ = ')';
258 				} else {
259 					struct tbl v;
260 					char *p;
261 
262 					v.flag = DEFINED|ISSET|INTEGER;
263 					v.type = 10; /* not default */
264 					v.name[0] = '\0';
265 					v_evaluate(&v, substitute(sp, 0),
266 						KSH_UNWIND_ERROR);
267 					sp = strchr(sp, 0) + 1;
268 					for (p = str_val(&v); *p; ) {
269 						Xcheck(ds, dp);
270 						*dp++ = *p++;
271 					}
272 				}
273 				continue;
274 			  case OSUBST: /* ${{#}var{:}[=+-?#%]word} */
275 			  /* format is:
276 			   *   OSUBST [{x] plain-variable-part \0
277 			   *     compiled-word-part CSUBST [}x]
278 			   * This is were all syntax checking gets done...
279 			   */
280 			  {
281 				char *varname = ++sp; /* skip the { or x (}) */
282 				int stype;
283 				int slen;
284 
285 				sp = strchr(sp, '\0') + 1; /* skip variable */
286 				type = varsub(&x, varname, sp, &stype, &slen);
287 				if (type < 0) {
288 					char endc;
289 					char *str, *end;
290 
291 					end = (char *) wdscan(sp, CSUBST);
292 					/* ({) the } or x is already skipped */
293 					endc = *end;
294 					*end = EOS;
295 					str = snptreef((char *) 0, 64, "%S",
296 							varname - 1);
297 					*end = endc;
298 					errorf("%s: bad substitution", str);
299 				}
300 				if (f&DOBLANK)
301 					doblank++;
302 				tilde_ok = 0;
303 				if (type == XBASE) {	/* expand? */
304 					if (!st->next) {
305 						SubType *newst;
306 
307 						newst = (SubType *) alloc(
308 							sizeof(SubType), ATEMP);
309 						newst->next = (SubType *) 0;
310 						newst->prev = st;
311 						st->next = newst;
312 					}
313 					st = st->next;
314 					st->stype = stype;
315 					st->base = Xsavepos(ds, dp);
316 					st->f = f;
317 					st->var = x.var;
318 					st->quote = quote;
319 					/* skip qualifier(s) */
320 					if (stype)
321 						sp += slen;
322 					switch (stype & 0x7f) {
323 					  case '#':
324 					  case '%':
325 						/* ! DOBLANK,DOBRACE_,DOTILDE */
326 						f = DOPAT | (f&DONTRUNCOMMAND)
327 						    | DOTEMP_;
328 						quote = 0;
329 						/* Prepend open pattern (so |
330 						 * in a trim will work as
331 						 * expected)
332 						 */
333 						*dp++ = MAGIC;
334 						*dp++ = '@' + 0x80;
335 						break;
336 					  case '=':
337 						/* Enabling tilde expansion
338 						 * after :'s here is
339 						 * non-standard ksh, but is
340 						 * consistent with rules for
341 						 * other assignments.  Not
342 						 * sure what POSIX thinks of
343 						 * this.
344 						 * Not doing tilde expansion
345 						 * for integer variables is a
346 						 * non-POSIX thing - makes
347 						 * sense though, since ~ is
348 						 * a arithmetic operator.
349 						 */
350 						if (!(x.var->flag & INTEGER))
351 							f |= DOASNTILDE|DOTILDE;
352 						f |= DOTEMP_;
353 						/* These will be done after the
354 						 * value has been assigned.
355 						 */
356 						f &= ~(DOBLANK|DOGLOB|DOBRACE_);
357 						tilde_ok = 1;
358 						break;
359 					  case '?':
360 						f &= ~DOBLANK;
361 						f |= DOTEMP_;
362 						/* fall through */
363 					  default:
364 						/* Enable tilde expansion */
365 						tilde_ok = 1;
366 						f |= DOTILDE;
367 					}
368 				} else
369 					/* skip word */
370 					sp = (char *) wdscan(sp, CSUBST);
371 				continue;
372 			  }
373 			  case CSUBST: /* only get here if expanding word */
374 				sp++; /* ({) skip the } or x */
375 				tilde_ok = 0;	/* in case of ${unset:-} */
376 				*dp = '\0';
377 				quote = st->quote;
378 				f = st->f;
379 				if (f&DOBLANK)
380 					doblank--;
381 				switch (st->stype&0x7f) {
382 				  case '#':
383 				  case '%':
384 					/* Append end-pattern */
385 					*dp++ = MAGIC; *dp++ = ')'; *dp = '\0';
386 					dp = Xrestpos(ds, dp, st->base);
387 					/* Must use st->var since calling
388 					 * global would break things
389 					 * like x[i+=1].
390 					 */
391 					x.str = trimsub(str_val(st->var),
392 						dp, st->stype);
393 					type = XSUB;
394 					if (f&DOBLANK)
395 						doblank++;
396 					st = st->prev;
397 					continue;
398 				  case '=':
399 					/* Restore our position and substitute
400 					 * the value of st->var (may not be
401 					 * the assigned value in the presence
402 					 * of integer/right-adj/etc attributes).
403 					 */
404 					dp = Xrestpos(ds, dp, st->base);
405 					/* Must use st->var since calling
406 					 * global would cause with things
407 					 * like x[i+=1] to be evaluated twice.
408 					 */
409 					/* Note: not exported by FEXPORT
410 					 * in at&t ksh.
411 					 */
412 					/* XXX POSIX says readonly is only
413 					 * fatal for special builtins (setstr
414 					 * does readonly check).
415 					 */
416 					setstr(st->var, debunk(
417 						(char *) alloc(strlen(dp) + 1,
418 							ATEMP), dp),
419 						KSH_UNWIND_ERROR);
420 					x.str = str_val(st->var);
421 					type = XSUB;
422 					if (f&DOBLANK)
423 						doblank++;
424 					st = st->prev;
425 					continue;
426 				  case '?':
427 				    {
428 					char *s = Xrestpos(ds, dp, st->base);
429 
430 					errorf("%s: %s", st->var->name,
431 					    dp == s ?
432 					      "parameter null or not set"
433 					    : (debunk(s, s), s));
434 				    }
435 				}
436 				st = st->prev;
437 				type = XBASE;
438 				continue;
439 
440 			  case OPAT: /* open pattern: *(foo|bar) */
441 				/* Next char is the type of pattern */
442 				make_magic = 1;
443 				c = *sp++ + 0x80;
444 				break;
445 
446 			  case SPAT: /* pattern seperator (|) */
447 				make_magic = 1;
448 				c = '|';
449 				break;
450 
451 			  case CPAT: /* close pattern */
452 				make_magic = 1;
453 				c = /*(*/ ')';
454 				break;
455 			}
456 			break;
457 
458 		  case XNULLSUB:
459 			/* Special case for "$@" (and "${foo[@]}") - no
460 			 * word is generated if $# is 0 (unless there is
461 			 * other stuff inside the quotes).
462 			 */
463 			type = XBASE;
464 			if (f&DOBLANK) {
465 				doblank--;
466 				/* not really correct: x=; "$x$@" should
467 				 * generate a null argument and
468 				 * set A; "${@:+}" shouldn't.
469 				 */
470 				if (dp == Xstring(ds, dp))
471 					word = IFS_WS;
472 			}
473 			continue;
474 
475 		  case XSUB:
476 			if ((c = *x.str++) == 0) {
477 				type = XBASE;
478 				if (f&DOBLANK)
479 					doblank--;
480 				continue;
481 			}
482 			break;
483 
484 		  case XARGSEP:
485 			type = XARG;
486 			quote = 1;
487 		  case XARG:
488 			if ((c = *x.str++) == '\0') {
489 				/* force null words to be created so
490 				 * set -- '' 2 ''; foo "$@" will do
491 				 * the right thing
492 				 */
493 				if (quote && x.split)
494 					word = IFS_WORD;
495 				if ((x.str = *x.u.strv++) == NULL) {
496 					type = XBASE;
497 					if (f&DOBLANK)
498 						doblank--;
499 					continue;
500 				}
501 				c = ifs0;
502 				if (c == 0) {
503 					if (quote && !x.split)
504 						continue;
505 					c = ' ';
506 				}
507 				if (quote && x.split) {
508 					/* terminate word for "$@" */
509 					type = XARGSEP;
510 					quote = 0;
511 				}
512 			}
513 			break;
514 
515 		  case XCOM:
516 			if (newlines) {		/* Spit out saved nl's */
517 				c = '\n';
518 				--newlines;
519 			} else {
520 				while ((c = shf_getc(x.u.shf)) == 0 || c == '\n')
521 				    if (c == '\n')
522 					    newlines++;	/* Save newlines */
523 				if (newlines && c != EOF) {
524 					shf_ungetc(c, x.u.shf);
525 					c = '\n';
526 					--newlines;
527 				}
528 			}
529 			if (c == EOF) {
530 				newlines = 0;
531 				shf_close(x.u.shf);
532 				if (x.split)
533 					subst_exstat = waitlast();
534 				type = XBASE;
535 				if (f&DOBLANK)
536 					doblank--;
537 				continue;
538 			}
539 			break;
540 		}
541 
542 		/* check for end of word or IFS separation */
543 		if (c == 0 || (!quote && (f & DOBLANK) && doblank && !make_magic
544 			       && ctype(c, C_IFS)))
545 		{
546 			/* How words are broken up:
547 			 *		   |       value of c
548 			 *	  word	   |	ws	nws	0
549 			 *	-----------------------------------
550 			 *	IFS_WORD	w/WS	w/NWS	w
551 			 *	IFS_WS		-/WS	w/NWS	-
552 			 *	IFS_NWS		-/NWS	w/NWS	w
553 			 *   (w means generate a word)
554 			 * Note that IFS_NWS/0 generates a word (at&t ksh
555 			 * doesn't do this, but POSIX does).
556 			 */
557 			if (word == IFS_WORD
558 			    || (!ctype(c, C_IFSWS) && (c || word == IFS_NWS)))
559 			{
560 				char *p;
561 
562 				*dp++ = '\0';
563 				p = Xclose(ds, dp);
564 #ifdef BRACE_EXPAND
565 				if (fdo & DOBRACE_)
566 					/* also does globbing */
567 					alt_expand(wp, p, p,
568 						   p + Xlength(ds, (dp - 1)),
569 						   fdo | (f & DOMARKDIRS));
570 				else
571 #endif /* BRACE_EXPAND */
572 				if (fdo & DOGLOB)
573 					glob(p, wp, f & DOMARKDIRS);
574 				else if ((f & DOPAT) || !(fdo & DOMAGIC_))
575 					XPput(*wp, p);
576 				else
577 					XPput(*wp, debunk(p, p));
578 				fdo = 0;
579 				saw_eq = 0;
580 				tilde_ok = (f & (DOTILDE|DOASNTILDE)) ? 1 : 0;
581 				if (c != 0)
582 					Xinit(ds, dp, 128, ATEMP);
583 			}
584 			if (c == 0)
585 				return;
586 			if (word != IFS_NWS)
587 				word = ctype(c, C_IFSWS) ? IFS_WS : IFS_NWS;
588 		} else {
589 			/* age tilde_ok info - ~ code tests second bit */
590 			tilde_ok <<= 1;
591 			/* mark any special second pass chars */
592 			if (!quote)
593 				switch (c) {
594 				  case '[':
595 				  case NOT:
596 				  case '-':
597 				  case ']':
598 					/* For character classes - doesn't hurt
599 					 * to have magic !,-,]'s outside of
600 					 * [...] expressions.
601 					 */
602 					if (f & (DOPAT | DOGLOB)) {
603 						fdo |= DOMAGIC_;
604 						if (c == '[')
605 							fdo |= f & DOGLOB;
606 						*dp++ = MAGIC;
607 					}
608 					break;
609 				  case '*':
610 				  case '?':
611 					if (f & (DOPAT | DOGLOB)) {
612 						fdo |= DOMAGIC_ | (f & DOGLOB);
613 						*dp++ = MAGIC;
614 					}
615 					break;
616 #ifdef BRACE_EXPAND
617 				  case OBRACE:
618 				  case ',':
619 				  case CBRACE:
620 					if ((f & DOBRACE_) && (c == OBRACE
621 						|| (fdo & DOBRACE_)))
622 					{
623 						fdo |= DOBRACE_|DOMAGIC_;
624 						*dp++ = MAGIC;
625 					}
626 					break;
627 #endif /* BRACE_EXPAND */
628 				  case '=':
629 					/* Note first unquoted = for ~ */
630 					if (!(f & DOTEMP_) && !saw_eq) {
631 						saw_eq = 1;
632 						tilde_ok = 1;
633 					}
634 					break;
635 				  case PATHSEP: /* : */
636 					/* Note unquoted : for ~ */
637 					if (!(f & DOTEMP_) && (f & DOASNTILDE))
638 						tilde_ok = 1;
639 					break;
640 				  case '~':
641 					/* tilde_ok is reset whenever
642 					 * any of ' " $( $(( ${ } are seen.
643 					 * Note that tilde_ok must be preserved
644 					 * through the sequence ${A=a=}~
645 					 */
646 					if (type == XBASE
647 					    && (f & (DOTILDE|DOASNTILDE))
648 					    && (tilde_ok & 2))
649 					{
650 						char *p, *dp_x;
651 
652 						dp_x = dp;
653 						p = maybe_expand_tilde(sp,
654 							&ds, &dp_x,
655 							f & DOASNTILDE);
656 						if (p) {
657 							if (dp != dp_x)
658 								word = IFS_WORD;
659 							dp = dp_x;
660 							sp = p;
661 							continue;
662 						}
663 					}
664 					break;
665 				}
666 			else
667 				quote &= ~2; /* undo temporary */
668 
669 			if (make_magic) {
670 				make_magic = 0;
671 				fdo |= DOMAGIC_ | (f & DOGLOB);
672 				*dp++ = MAGIC;
673 			} else if (ISMAGIC(c)) {
674 				fdo |= DOMAGIC_;
675 				*dp++ = MAGIC;
676 			}
677 			*dp++ = c; /* save output char */
678 			word = IFS_WORD;
679 		}
680 	}
681 }
682 
683 /*
684  * Prepare to generate the string returned by ${} substitution.
685  */
686 static int
varsub(xp,sp,word,stypep,slenp)687 varsub(xp, sp, word, stypep, slenp)
688 	Expand *xp;
689 	char *sp;
690 	char *word;
691 	int *stypep;	/* becomes qualifier type */
692 	int *slenp;	/* " " len (=, :=, etc.) valid iff *stypep != 0 */
693 {
694 	int c;
695 	int state;	/* next state: XBASE, XARG, XSUB, XNULLSUB */
696 	int stype;	/* substitution type */
697 	int slen;
698 	char *p;
699 	struct tbl *vp;
700 
701 	if (sp[0] == '\0')	/* Bad variable name */
702 		return -1;
703 
704 	xp->var = (struct tbl *) 0;
705 
706 	/* ${#var}, string length or array size */
707 	if (sp[0] == '#' && (c = sp[1]) != '\0') {
708 		int zero_ok = 0;
709 
710 		/* Can't have any modifiers for ${#...} */
711 		if (*word != CSUBST)
712 			return -1;
713 		sp++;
714 		/* Check for size of array */
715 		if ((p=strchr(sp,'[')) && (p[1]=='*'||p[1]=='@') && p[2]==']') {
716 			int n = 0;
717 			int max = 0;
718 			vp = global(arrayname(sp));
719 			if (vp->flag & (ISSET|ARRAY))
720 				zero_ok = 1;
721 			for (; vp; vp = vp->u.array)
722 				if (vp->flag & ISSET) {
723 					max = vp->index + 1;
724 					n++;
725 				}
726 			c = n; /* ksh88/ksh93 go for number, not max index */
727 		} else if (c == '*' || c == '@')
728 			c = e->loc->argc;
729 		else {
730 			p = str_val(global(sp));
731 			zero_ok = p != null;
732 			c = strlen(p);
733 		}
734 		if (Flag(FNOUNSET) && c == 0 && !zero_ok)
735 			errorf("%s: parameter not set", sp);
736 		*stypep = 0; /* unqualified variable/string substitution */
737 		xp->str = str_save(ulton((unsigned long)c, 10), ATEMP);
738 		return XSUB;
739 	}
740 
741 	/* Check for qualifiers in word part */
742 	stype = 0;
743 	c = word[slen = 0] == CHAR ? word[1] : 0;
744 	if (c == ':') {
745 		slen += 2;
746 		stype = 0x80;
747 		c = word[slen + 0] == CHAR ? word[slen + 1] : 0;
748 	}
749 	if (ctype(c, C_SUBOP1)) {
750 		slen += 2;
751 		stype |= c;
752 	} else if (ctype(c, C_SUBOP2)) { /* Note: ksh88 allows :%, :%%, etc */
753 		slen += 2;
754 		stype = c;
755 		if (word[slen + 0] == CHAR && c == word[slen + 1]) {
756 			stype |= 0x80;
757 			slen += 2;
758 		}
759 	} else if (stype)	/* : is not ok */
760 		return -1;
761 	if (!stype && *word != CSUBST)
762 		return -1;
763 	*stypep = stype;
764 	*slenp = slen;
765 
766 	c = sp[0];
767 	if (c == '*' || c == '@') {
768 		switch (stype & 0x7f) {
769 		  case '=':	/* can't assign to a vector */
770 		  case '%':	/* can't trim a vector (yet) */
771 		  case '#':
772 			return -1;
773 		}
774 		if (e->loc->argc == 0) {
775 			xp->str = null;
776 			state = c == '@' ? XNULLSUB : XSUB;
777 		} else {
778 			xp->u.strv = (const char **) e->loc->argv + 1;
779 			xp->str = *xp->u.strv++;
780 			xp->split = c == '@'; /* $@ */
781 			state = XARG;
782 		}
783 	} else {
784 		if ((p=strchr(sp,'[')) && (p[1]=='*'||p[1]=='@') && p[2]==']') {
785 			XPtrV wv;
786 
787 			switch (stype & 0x7f) {
788 			  case '=':	/* can't assign to a vector */
789 			  case '%':	/* can't trim a vector (yet) */
790 			  case '#':
791 				return -1;
792 			}
793 			XPinit(wv, 32);
794 			vp = global(arrayname(sp));
795 			for (; vp; vp = vp->u.array) {
796 				if (!(vp->flag&ISSET))
797 					continue;
798 				XPput(wv, str_val(vp));
799 			}
800 			if (XPsize(wv) == 0) {
801 				xp->str = null;
802 				state = p[1] == '@' ? XNULLSUB : XSUB;
803 				XPfree(wv);
804 			} else {
805 				XPput(wv, 0);
806 				xp->u.strv = (const char **) XPptrv(wv);
807 				xp->str = *xp->u.strv++;
808 				xp->split = p[1] == '@'; /* ${foo[@]} */
809 				state = XARG;
810 			}
811 		} else {
812 			/* Can't assign things like $! or $1 */
813 			if ((stype & 0x7f) == '='
814 			    && (ctype(*sp, C_VAR1) || digit(*sp)))
815 				return -1;
816 			xp->var = global(sp);
817 			xp->str = str_val(xp->var);
818 			state = XSUB;
819 		}
820 	}
821 
822 	c = stype&0x7f;
823 	/* test the compiler's code generator */
824 	if (ctype(c, C_SUBOP2) ||
825 	    (((stype&0x80) ? *xp->str=='\0' : xp->str==null) ? /* undef? */
826 	     c == '=' || c == '-' || c == '?' : c == '+'))
827 		state = XBASE;	/* expand word instead of variable value */
828 	if (Flag(FNOUNSET) && xp->str == null
829 	    && (ctype(c, C_SUBOP2) || (state != XBASE && c != '+')))
830 		errorf("%s: parameter not set", sp);
831 	return state;
832 }
833 
834 /*
835  * Run the command in $(...) and read its output.
836  */
837 static int
comsub(xp,cp)838 comsub(xp, cp)
839 	register Expand *xp;
840 	char *cp;
841 {
842 	Source *s, *sold;
843 	register struct op *t;
844 	struct shf *shf;
845 
846 	s = pushs(SSTRING, ATEMP);
847 	s->start = s->str = cp;
848 	sold = source;
849 	t = compile(s);
850 	source = sold;
851 
852 	if (t == NULL)
853 		return XBASE;
854 
855 	if (t != NULL && t->type == TCOM && /* $(<file) */
856 	    *t->args == NULL && *t->vars == NULL && t->ioact != NULL) {
857 		register struct ioword *io = *t->ioact;
858 		char *name;
859 
860 		if ((io->flag&IOTYPE) != IOREAD)
861 			errorf("funny $() command: %s",
862 				snptreef((char *) 0, 32, "%R", io));
863 		shf = shf_open(name = evalstr(io->name, DOTILDE), O_RDONLY, 0,
864 			SHF_MAPHI|SHF_CLEXEC);
865 		if (shf == NULL)
866 			errorf("%s: cannot open $() input", name);
867 		xp->split = 0;	/* no waitlast() */
868 	} else {
869 		int ofd1, pv[2];
870 		openpipe(pv);
871 		shf = shf_fdopen(pv[0], SHF_RD, (struct shf *) 0);
872 		ofd1 = savefd(1, 0);	/* fd 1 may be closed... */
873 		ksh_dup2(pv[1], 1, FALSE);
874 		close(pv[1]);
875 		execute(t, XFORK|XXCOM|XPIPEO);
876 		restfd(1, ofd1);
877 		startlast();
878 		xp->split = 1;	/* waitlast() */
879 	}
880 
881 	xp->u.shf = shf;
882 	return XCOM;
883 }
884 
885 /*
886  * perform #pattern and %pattern substitution in ${}
887  */
888 
889 static char *
trimsub(str,pat,how)890 trimsub(str, pat, how)
891 	register char *str;
892 	char *pat;
893 	int how;
894 {
895 	register char *end = strchr(str, 0);
896 	register char *p, c;
897 
898 	switch (how&0xff) {	/* UCHAR_MAX maybe? */
899 	  case '#':		/* shortest at begining */
900 		for (p = str; p <= end; p++) {
901 			c = *p; *p = '\0';
902 			if (gmatch(str, pat, FALSE)) {
903 				*p = c;
904 				return p;
905 			}
906 			*p = c;
907 		}
908 		break;
909 	  case '#'|0x80:	/* longest match at begining */
910 		for (p = end; p >= str; p--) {
911 			c = *p; *p = '\0';
912 			if (gmatch(str, pat, FALSE)) {
913 				*p = c;
914 				return p;
915 			}
916 			*p = c;
917 		}
918 		break;
919 	  case '%':		/* shortest match at end */
920 		for (p = end; p >= str; p--) {
921 			if (gmatch(p, pat, FALSE))
922 				return str_nsave(str, p - str, ATEMP);
923 		}
924 		break;
925 	  case '%'|0x80:	/* longest match at end */
926 		for (p = str; p <= end; p++) {
927 			if (gmatch(p, pat, FALSE))
928 				return str_nsave(str, p - str, ATEMP);
929 		}
930 		break;
931 	}
932 
933 	return str;		/* no match, return string */
934 }
935 
936 /*
937  * glob
938  * Name derived from V6's /etc/glob, the program that expanded filenames.
939  */
940 
941 /* XXX cp not const 'cause slashes are temporarily replaced with nulls... */
942 static void
glob(cp,wp,markdirs)943 glob(cp, wp, markdirs)
944 	char *cp;
945 	register XPtrV *wp;
946 	int markdirs;
947 {
948 	int oldsize = XPsize(*wp);
949 
950 	if (glob_str(cp, wp, markdirs) == 0)
951 		XPput(*wp, debunk(cp, cp));
952 	else
953 		qsortp(XPptrv(*wp) + oldsize, (size_t)(XPsize(*wp) - oldsize),
954 			xstrcmp);
955 }
956 
957 #define GF_NONE		0
958 #define GF_EXCHECK	BIT(0)		/* do existance check on file */
959 #define GF_GLOBBED	BIT(1)		/* some globbing has been done */
960 #define GF_MARKDIR	BIT(2)		/* add trailing / to directories */
961 
962 /* Apply file globbing to cp and store the matching files in wp.  Returns
963  * the number of matches found.
964  */
965 int
glob_str(cp,wp,markdirs)966 glob_str(cp, wp, markdirs)
967 	char *cp;
968 	XPtrV *wp;
969 	int markdirs;
970 {
971 	int oldsize = XPsize(*wp);
972 	XString xs;
973 	char *xp;
974 
975 	Xinit(xs, xp, 256, ATEMP);
976 	globit(&xs, &xp, cp, wp, markdirs ? GF_MARKDIR : GF_NONE);
977 	Xfree(xs, xp);
978 
979 	return XPsize(*wp) - oldsize;
980 }
981 
982 static void
globit(xs,xpp,sp,wp,check)983 globit(xs, xpp, sp, wp, check)
984 	XString *xs;		/* dest string */
985 	char **xpp;		/* ptr to dest end */
986 	char *sp;		/* source path */
987 	register XPtrV *wp;	/* output list */
988 	int check;		/* GF_* flags */
989 {
990 	register char *np;	/* next source component */
991 	char *xp = *xpp;
992 	char *se;
993 	char odirsep;
994 
995 	/* This to allow long expansions to be interrupted */
996 	intrcheck();
997 
998 	if (sp == NULL) {	/* end of source path */
999 		/* We only need to check if the file exists if a pattern
1000 		 * is followed by a non-pattern (eg, foo*x/bar; no check
1001 		 * is needed for foo* since the match must exist) or if
1002 		 * any patterns were expanded and the markdirs option is set.
1003 		 * Symlinks make things a bit tricky...
1004 		 */
1005 		if ((check & GF_EXCHECK)
1006 		    || ((check & GF_MARKDIR) && (check & GF_GLOBBED)))
1007 		{
1008 #define stat_check()	(stat_done ? stat_done : \
1009 			    (stat_done = stat(Xstring(*xs, xp), &statb) < 0 \
1010 				? -1 : 1))
1011 			struct stat lstatb, statb;
1012 			int stat_done = 0;	 /* -1: failed, 1 ok */
1013 
1014 			if (lstat(Xstring(*xs, xp), &lstatb) < 0)
1015 				return;
1016 			/* special case for systems which strip trailing
1017 			 * slashes from regular files (eg, /etc/passwd/).
1018 			 * SunOS 4.1.3 does this...
1019 			 */
1020 			if ((check & GF_EXCHECK) && xp > Xstring(*xs, xp)
1021 			    && ISDIRSEP(xp[-1]) && !S_ISDIR(lstatb.st_mode)
1022 #ifdef S_ISLNK
1023 			    && (!S_ISLNK(lstatb.st_mode)
1024 				|| stat_check() < 0
1025 				|| !S_ISDIR(statb.st_mode))
1026 #endif /* S_ISLNK */
1027 				)
1028 				return;
1029 			/* Possibly tack on a trailing / if there isn't already
1030 			 * one and if the file is a directory or a symlink to a
1031 			 * directory
1032 			 */
1033 			if (((check & GF_MARKDIR) && (check & GF_GLOBBED))
1034 			    && xp > Xstring(*xs, xp) && !ISDIRSEP(xp[-1])
1035 			    && (S_ISDIR(lstatb.st_mode)
1036 #ifdef S_ISLNK
1037 				|| (S_ISLNK(lstatb.st_mode)
1038 				    && stat_check() > 0
1039 				    && S_ISDIR(statb.st_mode))
1040 #endif /* S_ISLNK */
1041 				    ))
1042 			{
1043 				*xp++ = DIRSEP;
1044 				*xp = '\0';
1045 			}
1046 		}
1047 #ifdef OS2 /* Done this way to avoid bug in gcc 2.7.2... */
1048     /* Ugly kludge required for command
1049      * completion - see how search_access()
1050      * is implemented for OS/2...
1051      */
1052 # define KLUDGE_VAL	4
1053 #else /* OS2 */
1054 # define KLUDGE_VAL	0
1055 #endif /* OS2 */
1056 		XPput(*wp, str_nsave(Xstring(*xs, xp), Xlength(*xs, xp)
1057 			+ KLUDGE_VAL, ATEMP));
1058 		return;
1059 	}
1060 
1061 	if (xp > Xstring(*xs, xp))
1062 		*xp++ = DIRSEP;
1063 	while (ISDIRSEP(*sp)) {
1064 		Xcheck(*xs, xp);
1065 		*xp++ = *sp++;
1066 	}
1067 	np = ksh_strchr_dirsep(sp);
1068 	if (np != NULL) {
1069 		se = np;
1070 		odirsep = *np;	/* don't assume DIRSEP, can be multiple kinds */
1071 		*np++ = '\0';
1072 	} else {
1073 		odirsep = '\0'; /* keep gcc quiet */
1074 		se = sp + strlen(sp);
1075 	}
1076 
1077 
1078 	/* Check if sp needs globbing - done to avoid pattern checks for strings
1079 	 * containing MAGIC characters, open ['s without the matching close ],
1080 	 * etc. (otherwise opendir() will be called which may fail because the
1081 	 * directory isn't readable - if no globbing is needed, only execute
1082 	 * permission should be required (as per POSIX)).
1083 	 */
1084 	if (!has_globbing(sp, se)) {
1085 		XcheckN(*xs, xp, se - sp + 1);
1086 		debunk(xp, sp);
1087 		xp += strlen(xp);
1088 		*xpp = xp;
1089 		globit(xs, xpp, np, wp, check);
1090 	} else {
1091 		DIR *dirp;
1092 		struct dirent *d;
1093 		char *name;
1094 		int len;
1095 		int prefix_len;
1096 
1097 		/* xp = *xpp;	   copy_non_glob() may have re-alloc'd xs */
1098 		*xp = '\0';
1099 		prefix_len = Xlength(*xs, xp);
1100 		dirp = ksh_opendir(prefix_len ? Xstring(*xs, xp) : ".");
1101 		if (dirp == NULL)
1102 			goto Nodir;
1103 		while ((d = readdir(dirp)) != NULL) {
1104 			name = d->d_name;
1105 			if (name[0] == '.' &&
1106 			    (name[1] == 0 || (name[1] == '.' && name[2] == 0)))
1107 				continue; /* always ignore . and .. */
1108 			if ((*name == '.' && *sp != '.')
1109 			    || !gmatch(name, sp, TRUE))
1110 				continue;
1111 
1112 			len = NLENGTH(d) + 1;
1113 			XcheckN(*xs, xp, len);
1114 			memcpy(xp, name, len);
1115 			*xpp = xp + len - 1;
1116 			globit(xs, xpp, np, wp,
1117 				(check & GF_MARKDIR) | GF_GLOBBED
1118 				| (np ? GF_EXCHECK : GF_NONE));
1119 			xp = Xstring(*xs, xp) + prefix_len;
1120 		}
1121 		closedir(dirp);
1122 	  Nodir:;
1123 	}
1124 
1125 	if (np != NULL)
1126 		*--np = odirsep;
1127 }
1128 
1129 #if 0
1130 /* Check if p contains something that needs globbing; if it does, 0 is
1131  * returned; if not, p is copied into xs/xp after stripping any MAGICs
1132  */
1133 static int	copy_non_glob ARGS((XString *xs, char **xpp, char *p));
1134 static int
1135 copy_non_glob(xs, xpp, p)
1136 	XString *xs;
1137 	char **xpp;
1138 	char *p;
1139 {
1140 	char *xp;
1141 	int len = strlen(p);
1142 
1143 	XcheckN(*xs, *xpp, len);
1144 	xp = *xpp;
1145 	for (; *p; p++) {
1146 		if (ISMAGIC(*p)) {
1147 			int c = *++p;
1148 
1149 			if (c == '*' || c == '?')
1150 				return 0;
1151 			if (*p == '[') {
1152 				char *q = p + 1;
1153 
1154 				if (ISMAGIC(*q) && q[1] == NOT)
1155 					q += 2;
1156 				if (ISMAGIC(*q) && q[1] == ']')
1157 					q += 2;
1158 				for (; *q; q++)
1159 					if (ISMAGIC(*q) && *++q == ']')
1160 						return 0;
1161 				/* pass a literal [ through */
1162 			}
1163 			/* must be a MAGIC-MAGIC, or MAGIC-!, MAGIC--, etc. */
1164 		}
1165 		*xp++ = *p;
1166 	}
1167 	*xp = '\0';
1168 	*xpp = xp;
1169 	return 1;
1170 }
1171 #endif /* 0 */
1172 
1173 /* remove MAGIC from string */
1174 char *
debunk(dp,sp)1175 debunk(dp, sp)
1176 	char *dp;
1177 	const char *sp;
1178 {
1179 	char *d, *s;
1180 
1181 	if ((s = strchr(sp, MAGIC))) {
1182 		memcpy(dp, sp, s - sp);
1183 		for (d = dp + (s - sp); *s; s++)
1184 			if (!ISMAGIC(*s) || !(*++s & 0x80)
1185 			    || !strchr("*+?@! ", *s & 0x7f))
1186 				*d++ = *s;
1187 			else {
1188 				/* extended pattern operators: *+?@! */
1189 				if ((*s & 0x7f) != ' ')
1190 					*d++ = *s & 0x7f;
1191 				*d++ = '(';
1192 			}
1193 		*d = '\0';
1194 	} else if (dp != sp)
1195 		strcpy(dp, sp);
1196 	return dp;
1197 }
1198 
1199 /* Check if p is an unquoted name, possibly followed by a / or :.  If so
1200  * puts the expanded version in *dcp,dp and returns a pointer in p just
1201  * past the name, otherwise returns 0.
1202  */
1203 static char *
maybe_expand_tilde(p,dsp,dpp,isassign)1204 maybe_expand_tilde(p, dsp, dpp, isassign)
1205 	char *p;
1206 	XString *dsp;
1207 	char **dpp;
1208 	int isassign;
1209 {
1210 	XString ts;
1211 	char *dp = *dpp;
1212 	char *tp, *r;
1213 
1214 	Xinit(ts, tp, 16, ATEMP);
1215 	/* : only for DOASNTILDE form */
1216 	while (p[0] == CHAR && !ISDIRSEP(p[1])
1217 	       && (!isassign || p[1] != PATHSEP))
1218 	{
1219 		Xcheck(ts, tp);
1220 		*tp++ = p[1];
1221 		p += 2;
1222 	}
1223 	*tp = '\0';
1224 	r = (p[0] == EOS || p[0] == CHAR || p[0] == CSUBST) ? tilde(Xstring(ts, tp)) : (char *) 0;
1225 	Xfree(ts, tp);
1226 	if (r) {
1227 		while (*r) {
1228 			Xcheck(*dsp, dp);
1229 			if (ISMAGIC(*r))
1230 				*dp++ = MAGIC;
1231 			*dp++ = *r++;
1232 		}
1233 		*dpp = dp;
1234 		r = p;
1235 	}
1236 	return r;
1237 }
1238 
1239 /*
1240  * tilde expansion
1241  *
1242  * based on a version by Arnold Robbins
1243  */
1244 
1245 static char *
tilde(cp)1246 tilde(cp)
1247 	char *cp;
1248 {
1249 	char *dp;
1250 
1251 	if (cp[0] == '\0')
1252 		dp = str_val(global("HOME"));
1253 	else if (cp[0] == '+' && cp[1] == '\0')
1254 		dp = str_val(global("PWD"));
1255 	else if (cp[0] == '-' && cp[1] == '\0')
1256 		dp = str_val(global("OLDPWD"));
1257 	else
1258 		dp = homedir(cp);
1259 	/* If HOME, PWD or OLDPWD are not set, don't expand ~ */
1260 	if (dp == null)
1261 		dp = (char *) 0;
1262 	return dp;
1263 }
1264 
1265 /*
1266  * map userid to user's home directory.
1267  * note that 4.3's getpw adds more than 6K to the shell,
1268  * and the YP version probably adds much more.
1269  * we might consider our own version of getpwnam() to keep the size down.
1270  */
1271 
1272 static char *
homedir(name)1273 homedir(name)
1274 	char *name;
1275 {
1276 	register struct tbl *ap;
1277 
1278 	ap = tenter(&homedirs, name, hash(name));
1279 	if (!(ap->flag & ISSET)) {
1280 #ifdef OS2
1281 		/* No usernames in OS2 - punt */
1282 		return NULL;
1283 #else /* OS2 */
1284 		struct passwd *pw;
1285 
1286 		pw = getpwnam(name);
1287 		if (pw == NULL)
1288 			return NULL;
1289 		ap->val.s = str_save(pw->pw_dir, APERM);
1290 		ap->flag |= DEFINED|ISSET|ALLOC;
1291 #endif /* OS2 */
1292 	}
1293 	return ap->val.s;
1294 }
1295 
1296 #ifdef BRACE_EXPAND
1297 static void
alt_expand(wp,start,exp_start,end,fdo)1298 alt_expand(wp, start, exp_start, end, fdo)
1299 	XPtrV *wp;
1300 	char *start, *exp_start;
1301 	char *end;
1302 	int fdo;
1303 {
1304 	int UNINITIALIZED(count);
1305 	char *brace_start, *brace_end, *UNINITIALIZED(comma);
1306 	char *field_start;
1307 	char *p;
1308 
1309 	/* search for open brace */
1310 	for (p = exp_start; (p = strchr(p, MAGIC)) && p[1] != OBRACE; p += 2)
1311 		;
1312 	brace_start = p;
1313 
1314 	/* find matching close brace, if any */
1315 	if (p) {
1316 		comma = (char *) 0;
1317 		count = 1;
1318 		for (p += 2; *p && count; p++) {
1319 			if (ISMAGIC(*p)) {
1320 				if (*++p == OBRACE)
1321 					count++;
1322 				else if (*p == CBRACE)
1323 					--count;
1324 				else if (*p == ',' && count == 1)
1325 					comma = p;
1326 			}
1327 		}
1328 	}
1329 	/* no valid expansions... */
1330 	if (!p || count != 0) {
1331 		/* Note that given a{{b,c} we do not expand anything (this is
1332 		 * what at&t ksh does.  This may be changed to do the {b,c}
1333 		 * expansion. }
1334 		 */
1335 		if (fdo & DOGLOB)
1336 			glob(start, wp, fdo & DOMARKDIRS);
1337 		else
1338 			XPput(*wp, debunk(start, start));
1339 		return;
1340 	}
1341 	brace_end = p;
1342 	if (!comma) {
1343 		alt_expand(wp, start, brace_end, end, fdo);
1344 		return;
1345 	}
1346 
1347 	/* expand expression */
1348 	field_start = brace_start + 2;
1349 	count = 1;
1350 	for (p = brace_start + 2; p != brace_end; p++) {
1351 		if (ISMAGIC(*p)) {
1352 			if (*++p == OBRACE)
1353 				count++;
1354 			else if ((*p == CBRACE && --count == 0)
1355 				 || (*p == ',' && count == 1))
1356 			{
1357 				char *new;
1358 				int l1, l2, l3;
1359 
1360 				l1 = brace_start - start;
1361 				l2 = (p - 1) - field_start;
1362 				l3 = end - brace_end;
1363 				new = (char *) alloc(l1 + l2 + l3 + 1, ATEMP);
1364 				memcpy(new, start, l1);
1365 				memcpy(new + l1, field_start, l2);
1366 				memcpy(new + l1 + l2, brace_end, l3);
1367 				new[l1 + l2 + l3] = '\0';
1368 				alt_expand(wp, new, new + l1,
1369 					   new + l1 + l2 + l3, fdo);
1370 				field_start = p + 1;
1371 			}
1372 		}
1373 	}
1374 	return;
1375 }
1376 #endif /* BRACE_EXPAND */
1377