]> git.saurik.com Git - apple/shell_cmds.git/blob - sh/expand.c
shell_cmds-203.tar.gz
[apple/shell_cmds.git] / sh / expand.c
1 /*-
2 * Copyright (c) 1991, 1993
3 * The Regents of the University of California. All rights reserved.
4 * Copyright (c) 1997-2005
5 * Herbert Xu <herbert@gondor.apana.org.au>. All rights reserved.
6 * Copyright (c) 2010-2015
7 * Jilles Tjoelker <jilles@stack.nl>. All rights reserved.
8 *
9 * This code is derived from software contributed to Berkeley by
10 * Kenneth Almquist.
11 *
12 * Redistribution and use in source and binary forms, with or without
13 * modification, are permitted provided that the following conditions
14 * are met:
15 * 1. Redistributions of source code must retain the above copyright
16 * notice, this list of conditions and the following disclaimer.
17 * 2. Redistributions in binary form must reproduce the above copyright
18 * notice, this list of conditions and the following disclaimer in the
19 * documentation and/or other materials provided with the distribution.
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: head/bin/sh/expand.c 303586 2016-07-31 13:11:34Z jilles $");
44
45 #include <sys/types.h>
46 #include <sys/time.h>
47 #include <sys/stat.h>
48 #include <dirent.h>
49 #include <errno.h>
50 #include <inttypes.h>
51 #include <limits.h>
52 #include <pwd.h>
53 #include <stdio.h>
54 #include <stdlib.h>
55 #include <string.h>
56 #include <unistd.h>
57 #include <wchar.h>
58 #include <wctype.h>
59
60 /*
61 * Routines to expand arguments to commands. We have to deal with
62 * backquotes, shell variables, and file metacharacters.
63 */
64
65 #include "shell.h"
66 #include "main.h"
67 #include "nodes.h"
68 #include "eval.h"
69 #include "expand.h"
70 #include "syntax.h"
71 #include "parser.h"
72 #include "jobs.h"
73 #include "options.h"
74 #include "var.h"
75 #include "input.h"
76 #include "output.h"
77 #include "memalloc.h"
78 #include "error.h"
79 #include "mystring.h"
80 #include "arith.h"
81 #include "show.h"
82 #include "builtins.h"
83
84 enum wordstate { WORD_IDLE, WORD_WS_DELIMITED, WORD_QUOTEMARK };
85
86 struct worddest {
87 struct arglist *list;
88 enum wordstate state;
89 };
90
91 static char *expdest; /* output of current string */
92 static struct nodelist *argbackq; /* list of back quote expressions */
93
94 static const char *argstr(const char *, int, struct worddest *);
95 static const char *exptilde(const char *, int);
96 static const char *expari(const char *, int, struct worddest *);
97 static void expbackq(union node *, int, int, struct worddest *);
98 static void subevalvar_trim(const char *, int, int, int);
99 static int subevalvar_misc(const char *, const char *, int, int, int);
100 static const char *evalvar(const char *, int, struct worddest *);
101 static int varisset(const char *, int);
102 static void strtodest(const char *, int, int, int, struct worddest *);
103 static void reprocess(int, int, int, int, struct worddest *);
104 static void varvalue(const char *, int, int, int, struct worddest *);
105 static void expandmeta(char *, struct arglist *);
106 static void expmeta(char *, char *, struct arglist *);
107 static int expsortcmp(const void *, const void *);
108 static int patmatch(const char *, const char *);
109 static void cvtnum(int, char *);
110 static int collate_range_cmp(wchar_t, wchar_t);
111
112 void
113 emptyarglist(struct arglist *list)
114 {
115
116 list->args = list->smallarg;
117 list->count = 0;
118 list->capacity = sizeof(list->smallarg) / sizeof(list->smallarg[0]);
119 }
120
121 void
122 appendarglist(struct arglist *list, char *str)
123 {
124 char **newargs;
125 int newcapacity;
126
127 if (list->count >= list->capacity) {
128 newcapacity = list->capacity * 2;
129 if (newcapacity < 16)
130 newcapacity = 16;
131 if (newcapacity > INT_MAX / (int)sizeof(newargs[0]))
132 error("Too many entries in arglist");
133 newargs = stalloc(newcapacity * sizeof(newargs[0]));
134 memcpy(newargs, list->args, list->count * sizeof(newargs[0]));
135 list->args = newargs;
136 list->capacity = newcapacity;
137 }
138 list->args[list->count++] = str;
139 }
140
141 static int
142 collate_range_cmp(wchar_t c1, wchar_t c2)
143 {
144 static wchar_t s1[2], s2[2];
145
146 s1[0] = c1;
147 s2[0] = c2;
148 return (wcscoll(s1, s2));
149 }
150
151 static char *
152 stputs_quotes(const char *data, const char *syntax, char *p)
153 {
154 while (*data) {
155 CHECKSTRSPACE(2, p);
156 if (syntax[(int)*data] == CCTL)
157 USTPUTC(CTLESC, p);
158 USTPUTC(*data++, p);
159 }
160 return (p);
161 }
162 #define STPUTS_QUOTES(data, syntax, p) p = stputs_quotes((data), syntax, p)
163
164 static char *
165 nextword(char c, int flag, char *p, struct worddest *dst)
166 {
167 int is_ws;
168
169 is_ws = c == '\t' || c == '\n' || c == ' ';
170 if (p != stackblock() || (is_ws ? dst->state == WORD_QUOTEMARK :
171 dst->state != WORD_WS_DELIMITED) || c == '\0') {
172 STPUTC('\0', p);
173 if (flag & EXP_GLOB)
174 expandmeta(grabstackstr(p), dst->list);
175 else
176 appendarglist(dst->list, grabstackstr(p));
177 dst->state = is_ws ? WORD_WS_DELIMITED : WORD_IDLE;
178 } else if (!is_ws && dst->state == WORD_WS_DELIMITED)
179 dst->state = WORD_IDLE;
180 /* Reserve space while the stack string is empty. */
181 appendarglist(dst->list, NULL);
182 dst->list->count--;
183 STARTSTACKSTR(p);
184 return p;
185 }
186 #define NEXTWORD(c, flag, p, dstlist) p = nextword(c, flag, p, dstlist)
187
188 static char *
189 stputs_split(const char *data, const char *syntax, int flag, char *p,
190 struct worddest *dst)
191 {
192 const char *ifs;
193 char c;
194
195 ifs = ifsset() ? ifsval() : " \t\n";
196 while (*data) {
197 CHECKSTRSPACE(2, p);
198 c = *data++;
199 if (strchr(ifs, c) != NULL) {
200 NEXTWORD(c, flag, p, dst);
201 continue;
202 }
203 if (flag & EXP_GLOB && syntax[(int)c] == CCTL)
204 USTPUTC(CTLESC, p);
205 USTPUTC(c, p);
206 }
207 return (p);
208 }
209 #define STPUTS_SPLIT(data, syntax, flag, p, dst) p = stputs_split((data), syntax, flag, p, dst)
210
211 /*
212 * Perform expansions on an argument, placing the resulting list of arguments
213 * in arglist. Parameter expansion, command substitution and arithmetic
214 * expansion are always performed; additional expansions can be requested
215 * via flag (EXP_*).
216 * The result is left in the stack string.
217 * When arglist is NULL, perform here document expansion.
218 *
219 * Caution: this function uses global state and is not reentrant.
220 * However, a new invocation after an interrupted invocation is safe
221 * and will reset the global state for the new call.
222 */
223 void
224 expandarg(union node *arg, struct arglist *arglist, int flag)
225 {
226 struct worddest exparg;
227
228 if (fflag)
229 flag &= ~EXP_GLOB;
230 argbackq = arg->narg.backquote;
231 exparg.list = arglist;
232 exparg.state = WORD_IDLE;
233 STARTSTACKSTR(expdest);
234 argstr(arg->narg.text, flag, &exparg);
235 if (arglist == NULL) {
236 STACKSTRNUL(expdest);
237 return; /* here document expanded */
238 }
239 if ((flag & EXP_SPLIT) == 0 || expdest != stackblock() ||
240 exparg.state == WORD_QUOTEMARK) {
241 STPUTC('\0', expdest);
242 if (flag & EXP_SPLIT) {
243 if (flag & EXP_GLOB)
244 expandmeta(grabstackstr(expdest), exparg.list);
245 else
246 appendarglist(exparg.list, grabstackstr(expdest));
247 }
248 }
249 if ((flag & EXP_SPLIT) == 0)
250 appendarglist(arglist, grabstackstr(expdest));
251 }
252
253
254
255 /*
256 * Perform parameter expansion, command substitution and arithmetic
257 * expansion, and tilde expansion if requested via EXP_TILDE/EXP_VARTILDE.
258 * Processing ends at a CTLENDVAR or CTLENDARI character as well as '\0'.
259 * This is used to expand word in ${var+word} etc.
260 * If EXP_GLOB or EXP_CASE are set, keep and/or generate CTLESC
261 * characters to allow for further processing.
262 *
263 * If EXP_SPLIT is set, dst receives any complete words produced.
264 */
265 static const char *
266 argstr(const char *p, int flag, struct worddest *dst)
267 {
268 char c;
269 int quotes = flag & (EXP_GLOB | EXP_CASE); /* do CTLESC */
270 int firsteq = 1;
271 int split_lit;
272 int lit_quoted;
273
274 split_lit = flag & EXP_SPLIT_LIT;
275 lit_quoted = flag & EXP_LIT_QUOTED;
276 flag &= ~(EXP_SPLIT_LIT | EXP_LIT_QUOTED);
277 if (*p == '~' && (flag & (EXP_TILDE | EXP_VARTILDE)))
278 p = exptilde(p, flag);
279 for (;;) {
280 CHECKSTRSPACE(2, expdest);
281 switch (c = *p++) {
282 case '\0':
283 return (p - 1);
284 case CTLENDVAR:
285 case CTLENDARI:
286 return (p);
287 case CTLQUOTEMARK:
288 lit_quoted = 1;
289 /* "$@" syntax adherence hack */
290 if (p[0] == CTLVAR && (p[1] & VSQUOTE) != 0 &&
291 p[2] == '@' && p[3] == '=')
292 break;
293 if ((flag & EXP_SPLIT) != 0 && expdest == stackblock())
294 dst->state = WORD_QUOTEMARK;
295 break;
296 case CTLQUOTEEND:
297 lit_quoted = 0;
298 break;
299 case CTLESC:
300 c = *p++;
301 if (split_lit && !lit_quoted &&
302 strchr(ifsset() ? ifsval() : " \t\n", c) != NULL) {
303 NEXTWORD(c, flag, expdest, dst);
304 break;
305 }
306 if (quotes)
307 USTPUTC(CTLESC, expdest);
308 USTPUTC(c, expdest);
309 break;
310 case CTLVAR:
311 p = evalvar(p, flag, dst);
312 break;
313 case CTLBACKQ:
314 case CTLBACKQ|CTLQUOTE:
315 expbackq(argbackq->n, c & CTLQUOTE, flag, dst);
316 argbackq = argbackq->next;
317 break;
318 case CTLARI:
319 p = expari(p, flag, dst);
320 break;
321 case ':':
322 case '=':
323 /*
324 * sort of a hack - expand tildes in variable
325 * assignments (after the first '=' and after ':'s).
326 */
327 if (split_lit && !lit_quoted &&
328 strchr(ifsset() ? ifsval() : " \t\n", c) != NULL) {
329 NEXTWORD(c, flag, expdest, dst);
330 break;
331 }
332 USTPUTC(c, expdest);
333 if (flag & EXP_VARTILDE && *p == '~' &&
334 (c != '=' || firsteq)) {
335 if (c == '=')
336 firsteq = 0;
337 p = exptilde(p, flag);
338 }
339 break;
340 default:
341 if (split_lit && !lit_quoted &&
342 strchr(ifsset() ? ifsval() : " \t\n", c) != NULL) {
343 NEXTWORD(c, flag, expdest, dst);
344 break;
345 }
346 USTPUTC(c, expdest);
347 }
348 }
349 }
350
351 /*
352 * Perform tilde expansion, placing the result in the stack string and
353 * returning the next position in the input string to process.
354 */
355 static const char *
356 exptilde(const char *p, int flag)
357 {
358 char c;
359 const char *startp = p;
360 const char *user;
361 struct passwd *pw;
362 char *home;
363 int len;
364
365 for (;;) {
366 c = *p;
367 switch(c) {
368 case CTLESC: /* This means CTL* are always considered quoted. */
369 case CTLVAR:
370 case CTLBACKQ:
371 case CTLBACKQ | CTLQUOTE:
372 case CTLARI:
373 case CTLENDARI:
374 case CTLQUOTEMARK:
375 return (startp);
376 case ':':
377 if ((flag & EXP_VARTILDE) == 0)
378 break;
379 /* FALLTHROUGH */
380 case '\0':
381 case '/':
382 case CTLENDVAR:
383 len = p - startp - 1;
384 STPUTBIN(startp + 1, len, expdest);
385 STACKSTRNUL(expdest);
386 user = expdest - len;
387 if (*user == '\0') {
388 home = lookupvar("HOME");
389 } else {
390 pw = getpwnam(user);
391 home = pw != NULL ? pw->pw_dir : NULL;
392 }
393 STADJUST(-len, expdest);
394 if (home == NULL || *home == '\0')
395 return (startp);
396 strtodest(home, flag, VSNORMAL, 1, NULL);
397 return (p);
398 }
399 p++;
400 }
401 }
402
403
404 /*
405 * Expand arithmetic expression.
406 */
407 static const char *
408 expari(const char *p, int flag, struct worddest *dst)
409 {
410 char *q, *start;
411 arith_t result;
412 int begoff;
413 int quoted;
414 int adj;
415
416 quoted = *p++ == '"';
417 begoff = expdest - stackblock();
418 p = argstr(p, 0, NULL);
419 STPUTC('\0', expdest);
420 start = stackblock() + begoff;
421
422 q = grabstackstr(expdest);
423 result = arith(start);
424 ungrabstackstr(q, expdest);
425
426 start = stackblock() + begoff;
427 adj = start - expdest;
428 STADJUST(adj, expdest);
429
430 CHECKSTRSPACE((int)(DIGITS(result) + 1), expdest);
431 fmtstr(expdest, DIGITS(result), ARITH_FORMAT_STR, result);
432 adj = strlen(expdest);
433 STADJUST(adj, expdest);
434 if (!quoted)
435 reprocess(expdest - adj - stackblock(), flag, VSNORMAL, 0, dst);
436 return p;
437 }
438
439
440 /*
441 * Perform command substitution.
442 */
443 static void
444 expbackq(union node *cmd, int quoted, int flag, struct worddest *dst)
445 {
446 struct backcmd in;
447 int i;
448 char buf[128];
449 char *p;
450 char *dest = expdest;
451 struct nodelist *saveargbackq;
452 char lastc;
453 char const *syntax = quoted? DQSYNTAX : BASESYNTAX;
454 int quotes = flag & (EXP_GLOB | EXP_CASE);
455 size_t nnl;
456 const char *ifs;
457
458 INTOFF;
459 saveargbackq = argbackq;
460 p = grabstackstr(dest);
461 evalbackcmd(cmd, &in);
462 ungrabstackstr(p, dest);
463 argbackq = saveargbackq;
464
465 p = in.buf;
466 nnl = 0;
467 if (!quoted && flag & EXP_SPLIT)
468 ifs = ifsset() ? ifsval() : " \t\n";
469 else
470 ifs = "";
471 /* Don't copy trailing newlines */
472 for (;;) {
473 if (--in.nleft < 0) {
474 if (in.fd < 0)
475 break;
476 while ((i = read(in.fd, buf, sizeof buf)) < 0 && errno == EINTR)
477 ;
478 TRACE(("expbackq: read returns %d\n", i));
479 if (i <= 0)
480 break;
481 p = buf;
482 in.nleft = i - 1;
483 }
484 lastc = *p++;
485 if (lastc == '\0')
486 continue;
487 if (lastc == '\n') {
488 nnl++;
489 } else {
490 if (nnl > 0) {
491 if (strchr(ifs, '\n') != NULL) {
492 NEXTWORD('\n', flag, dest, dst);
493 nnl = 0;
494 } else {
495 CHECKSTRSPACE(nnl + 2, dest);
496 while (nnl > 0) {
497 nnl--;
498 USTPUTC('\n', dest);
499 }
500 }
501 }
502 if (strchr(ifs, lastc) != NULL)
503 NEXTWORD(lastc, flag, dest, dst);
504 else {
505 CHECKSTRSPACE(2, dest);
506 if (quotes && syntax[(int)lastc] == CCTL)
507 USTPUTC(CTLESC, dest);
508 USTPUTC(lastc, dest);
509 }
510 }
511 }
512
513 if (in.fd >= 0)
514 close(in.fd);
515 if (in.buf)
516 ckfree(in.buf);
517 if (in.jp)
518 exitstatus = waitforjob(in.jp, (int *)NULL);
519 TRACE(("expbackq: size=%td: \"%.*s\"\n",
520 ((dest - stackblock()) - startloc),
521 (int)((dest - stackblock()) - startloc),
522 stackblock() + startloc));
523 expdest = dest;
524 INTON;
525 }
526
527
528
529 static void
530 recordleft(const char *str, const char *loc, char *startp)
531 {
532 int amount;
533
534 amount = ((str - 1) - (loc - startp)) - expdest;
535 STADJUST(amount, expdest);
536 while (loc != str - 1)
537 *startp++ = *loc++;
538 }
539
540 static void
541 subevalvar_trim(const char *p, int strloc, int subtype, int startloc)
542 {
543 char *startp;
544 char *loc = NULL;
545 char *str;
546 int c = 0;
547 struct nodelist *saveargbackq = argbackq;
548 int amount;
549
550 argstr(p, EXP_CASE | EXP_TILDE, NULL);
551 STACKSTRNUL(expdest);
552 argbackq = saveargbackq;
553 startp = stackblock() + startloc;
554 str = stackblock() + strloc;
555
556 switch (subtype) {
557 case VSTRIMLEFT:
558 for (loc = startp; loc < str; loc++) {
559 c = *loc;
560 *loc = '\0';
561 if (patmatch(str, startp)) {
562 *loc = c;
563 recordleft(str, loc, startp);
564 return;
565 }
566 *loc = c;
567 }
568 break;
569
570 case VSTRIMLEFTMAX:
571 for (loc = str - 1; loc >= startp;) {
572 c = *loc;
573 *loc = '\0';
574 if (patmatch(str, startp)) {
575 *loc = c;
576 recordleft(str, loc, startp);
577 return;
578 }
579 *loc = c;
580 loc--;
581 }
582 break;
583
584 case VSTRIMRIGHT:
585 for (loc = str - 1; loc >= startp;) {
586 if (patmatch(str, loc)) {
587 amount = loc - expdest;
588 STADJUST(amount, expdest);
589 return;
590 }
591 loc--;
592 }
593 break;
594
595 case VSTRIMRIGHTMAX:
596 for (loc = startp; loc < str - 1; loc++) {
597 if (patmatch(str, loc)) {
598 amount = loc - expdest;
599 STADJUST(amount, expdest);
600 return;
601 }
602 }
603 break;
604
605
606 default:
607 abort();
608 }
609 amount = (expdest - stackblock() - strloc) + 1;
610 STADJUST(-amount, expdest);
611 }
612
613
614 static int
615 subevalvar_misc(const char *p, const char *var, int subtype, int startloc,
616 int varflags)
617 {
618 char *startp;
619 struct nodelist *saveargbackq = argbackq;
620 int amount;
621
622 argstr(p, EXP_TILDE, NULL);
623 STACKSTRNUL(expdest);
624 argbackq = saveargbackq;
625 startp = stackblock() + startloc;
626
627 switch (subtype) {
628 case VSASSIGN:
629 setvar(var, startp, 0);
630 amount = startp - expdest;
631 STADJUST(amount, expdest);
632 return 1;
633
634 case VSQUESTION:
635 if (*p != CTLENDVAR) {
636 outfmt(out2, "%s\n", startp);
637 error((char *)NULL);
638 }
639 error("%.*s: parameter %snot set", (int)(p - var - 1),
640 var, (varflags & VSNUL) ? "null or " : "");
641 return 0;
642
643 default:
644 abort();
645 }
646 }
647
648
649 /*
650 * Expand a variable, and return a pointer to the next character in the
651 * input string.
652 */
653
654 static const char *
655 evalvar(const char *p, int flag, struct worddest *dst)
656 {
657 int subtype;
658 int varflags;
659 const char *var;
660 const char *val;
661 int patloc;
662 int c;
663 int set;
664 int special;
665 int startloc;
666 int varlen;
667 int varlenb;
668 char buf[21];
669
670 varflags = (unsigned char)*p++;
671 subtype = varflags & VSTYPE;
672 var = p;
673 special = 0;
674 if (! is_name(*p))
675 special = 1;
676 p = strchr(p, '=') + 1;
677 again: /* jump here after setting a variable with ${var=text} */
678 if (varflags & VSLINENO) {
679 set = 1;
680 special = 1;
681 val = NULL;
682 } else if (special) {
683 set = varisset(var, varflags & VSNUL);
684 val = NULL;
685 } else {
686 val = bltinlookup(var, 1);
687 if (val == NULL || ((varflags & VSNUL) && val[0] == '\0')) {
688 val = NULL;
689 set = 0;
690 } else
691 set = 1;
692 }
693 varlen = 0;
694 startloc = expdest - stackblock();
695 if (!set && uflag && *var != '@' && *var != '*') {
696 switch (subtype) {
697 case VSNORMAL:
698 case VSTRIMLEFT:
699 case VSTRIMLEFTMAX:
700 case VSTRIMRIGHT:
701 case VSTRIMRIGHTMAX:
702 case VSLENGTH:
703 error("%.*s: parameter not set", (int)(p - var - 1),
704 var);
705 }
706 }
707 if (set && subtype != VSPLUS) {
708 /* insert the value of the variable */
709 if (special) {
710 if (varflags & VSLINENO) {
711 if (p - var > (ptrdiff_t)sizeof(buf))
712 abort();
713 memcpy(buf, var, p - var - 1);
714 buf[p - var - 1] = '\0';
715 strtodest(buf, flag, subtype,
716 varflags & VSQUOTE, dst);
717 } else
718 varvalue(var, varflags & VSQUOTE, subtype, flag,
719 dst);
720 if (subtype == VSLENGTH) {
721 varlenb = expdest - stackblock() - startloc;
722 varlen = varlenb;
723 if (localeisutf8) {
724 val = stackblock() + startloc;
725 for (;val != expdest; val++)
726 if ((*val & 0xC0) == 0x80)
727 varlen--;
728 }
729 STADJUST(-varlenb, expdest);
730 }
731 } else {
732 if (subtype == VSLENGTH) {
733 for (;*val; val++)
734 if (!localeisutf8 ||
735 (*val & 0xC0) != 0x80)
736 varlen++;
737 }
738 else
739 strtodest(val, flag, subtype,
740 varflags & VSQUOTE, dst);
741 }
742 }
743
744 if (subtype == VSPLUS)
745 set = ! set;
746
747 switch (subtype) {
748 case VSLENGTH:
749 cvtnum(varlen, buf);
750 strtodest(buf, flag, VSNORMAL, varflags & VSQUOTE, dst);
751 break;
752
753 case VSNORMAL:
754 break;
755
756 case VSPLUS:
757 case VSMINUS:
758 if (!set) {
759 argstr(p, flag | (flag & EXP_SPLIT ? EXP_SPLIT_LIT : 0) |
760 (varflags & VSQUOTE ? EXP_LIT_QUOTED : 0), dst);
761 break;
762 }
763 break;
764
765 case VSTRIMLEFT:
766 case VSTRIMLEFTMAX:
767 case VSTRIMRIGHT:
768 case VSTRIMRIGHTMAX:
769 if (!set)
770 break;
771 /*
772 * Terminate the string and start recording the pattern
773 * right after it
774 */
775 STPUTC('\0', expdest);
776 patloc = expdest - stackblock();
777 subevalvar_trim(p, patloc, subtype, startloc);
778 reprocess(startloc, flag, VSNORMAL, varflags & VSQUOTE, dst);
779 if (flag & EXP_SPLIT && *var == '@' && varflags & VSQUOTE)
780 dst->state = WORD_QUOTEMARK;
781 break;
782
783 case VSASSIGN:
784 case VSQUESTION:
785 if (!set) {
786 if (subevalvar_misc(p, var, subtype, startloc,
787 varflags)) {
788 varflags &= ~VSNUL;
789 goto again;
790 }
791 break;
792 }
793 break;
794
795 case VSERROR:
796 c = p - var - 1;
797 error("${%.*s%s}: Bad substitution", c, var,
798 (c > 0 && *p != CTLENDVAR) ? "..." : "");
799
800 default:
801 abort();
802 }
803
804 if (subtype != VSNORMAL) { /* skip to end of alternative */
805 int nesting = 1;
806 for (;;) {
807 if ((c = *p++) == CTLESC)
808 p++;
809 else if (c == CTLBACKQ || c == (CTLBACKQ|CTLQUOTE)) {
810 if (set)
811 argbackq = argbackq->next;
812 } else if (c == CTLVAR) {
813 if ((*p++ & VSTYPE) != VSNORMAL)
814 nesting++;
815 } else if (c == CTLENDVAR) {
816 if (--nesting == 0)
817 break;
818 }
819 }
820 }
821 return p;
822 }
823
824
825
826 /*
827 * Test whether a specialized variable is set.
828 */
829
830 static int
831 varisset(const char *name, int nulok)
832 {
833
834 if (*name == '!')
835 return backgndpidset();
836 else if (*name == '@' || *name == '*') {
837 if (*shellparam.p == NULL)
838 return 0;
839
840 if (nulok) {
841 char **av;
842
843 for (av = shellparam.p; *av; av++)
844 if (**av != '\0')
845 return 1;
846 return 0;
847 }
848 } else if (is_digit(*name)) {
849 char *ap;
850 long num;
851
852 errno = 0;
853 num = strtol(name, NULL, 10);
854 if (errno != 0 || num > shellparam.nparam)
855 return 0;
856
857 if (num == 0)
858 ap = arg0;
859 else
860 ap = shellparam.p[num - 1];
861
862 if (nulok && (ap == NULL || *ap == '\0'))
863 return 0;
864 }
865 return 1;
866 }
867
868 static void
869 strtodest(const char *p, int flag, int subtype, int quoted,
870 struct worddest *dst)
871 {
872 if (subtype == VSLENGTH || subtype == VSTRIMLEFT ||
873 subtype == VSTRIMLEFTMAX || subtype == VSTRIMRIGHT ||
874 subtype == VSTRIMRIGHTMAX)
875 STPUTS(p, expdest);
876 else if (flag & EXP_SPLIT && !quoted && dst != NULL)
877 STPUTS_SPLIT(p, BASESYNTAX, flag, expdest, dst);
878 else if (flag & (EXP_GLOB | EXP_CASE))
879 STPUTS_QUOTES(p, quoted ? DQSYNTAX : BASESYNTAX, expdest);
880 else
881 STPUTS(p, expdest);
882 }
883
884 static void
885 reprocess(int startloc, int flag, int subtype, int quoted,
886 struct worddest *dst)
887 {
888 static char *buf = NULL;
889 static size_t buflen = 0;
890 char *startp;
891 size_t len, zpos, zlen;
892
893 startp = stackblock() + startloc;
894 len = expdest - startp;
895 if (len >= SIZE_MAX / 2)
896 abort();
897 INTOFF;
898 if (len >= buflen) {
899 ckfree(buf);
900 buf = NULL;
901 }
902 if (buflen < 128)
903 buflen = 128;
904 while (len >= buflen)
905 buflen <<= 1;
906 if (buf == NULL)
907 buf = ckmalloc(buflen);
908 INTON;
909 memcpy(buf, startp, len);
910 buf[len] = '\0';
911 STADJUST(-len, expdest);
912 for (zpos = 0;;) {
913 zlen = strlen(buf + zpos);
914 strtodest(buf + zpos, flag, subtype, quoted, dst);
915 zpos += zlen + 1;
916 if (zpos == len + 1)
917 break;
918 if (flag & EXP_SPLIT && (quoted || (zlen > 0 && zpos < len)))
919 NEXTWORD('\0', flag, expdest, dst);
920 }
921 }
922
923 /*
924 * Add the value of a specialized variable to the stack string.
925 */
926
927 static void
928 varvalue(const char *name, int quoted, int subtype, int flag,
929 struct worddest *dst)
930 {
931 int num;
932 char *p;
933 int i;
934 int splitlater;
935 char sep[2];
936 char **ap;
937 char buf[(NSHORTOPTS > 10 ? NSHORTOPTS : 10) + 1];
938
939 if (subtype == VSLENGTH)
940 flag &= ~EXP_FULL;
941 splitlater = subtype == VSTRIMLEFT || subtype == VSTRIMLEFTMAX ||
942 subtype == VSTRIMRIGHT || subtype == VSTRIMRIGHTMAX;
943
944 switch (*name) {
945 case '$':
946 num = rootpid;
947 break;
948 case '?':
949 num = oexitstatus;
950 break;
951 case '#':
952 num = shellparam.nparam;
953 break;
954 case '!':
955 num = backgndpidval();
956 break;
957 case '-':
958 p = buf;
959 for (i = 0 ; i < NSHORTOPTS ; i++) {
960 if (optval[i])
961 *p++ = optletter[i];
962 }
963 *p = '\0';
964 strtodest(buf, flag, subtype, quoted, dst);
965 return;
966 case '@':
967 if (flag & EXP_SPLIT && quoted) {
968 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) {
969 strtodest(p, flag, subtype, quoted, dst);
970 if (*ap) {
971 if (splitlater)
972 STPUTC('\0', expdest);
973 else
974 NEXTWORD('\0', flag, expdest,
975 dst);
976 }
977 }
978 if (shellparam.nparam > 0)
979 dst->state = WORD_QUOTEMARK;
980 return;
981 }
982 /* FALLTHROUGH */
983 case '*':
984 if (ifsset())
985 sep[0] = ifsval()[0];
986 else
987 sep[0] = ' ';
988 sep[1] = '\0';
989 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) {
990 strtodest(p, flag, subtype, quoted, dst);
991 if (!*ap)
992 break;
993 if (sep[0])
994 strtodest(sep, flag, subtype, quoted, dst);
995 else if (flag & EXP_SPLIT && !quoted && **ap != '\0') {
996 if (splitlater)
997 STPUTC('\0', expdest);
998 else
999 NEXTWORD('\0', flag, expdest, dst);
1000 }
1001 }
1002 return;
1003 default:
1004 if (is_digit(*name)) {
1005 num = atoi(name);
1006 if (num == 0)
1007 p = arg0;
1008 else if (num > 0 && num <= shellparam.nparam)
1009 p = shellparam.p[num - 1];
1010 else
1011 return;
1012 strtodest(p, flag, subtype, quoted, dst);
1013 }
1014 return;
1015 }
1016 cvtnum(num, buf);
1017 strtodest(buf, flag, subtype, quoted, dst);
1018 }
1019
1020
1021
1022 static char expdir[PATH_MAX];
1023 #define expdir_end (expdir + sizeof(expdir))
1024
1025 /*
1026 * Perform pathname generation and remove control characters.
1027 * At this point, the only control characters should be CTLESC.
1028 * The results are stored in the list dstlist.
1029 */
1030 static void
1031 expandmeta(char *pattern, struct arglist *dstlist)
1032 {
1033 char *p;
1034 int firstmatch;
1035 char c;
1036
1037 firstmatch = dstlist->count;
1038 p = pattern;
1039 for (; (c = *p) != '\0'; p++) {
1040 /* fast check for meta chars */
1041 if (c == '*' || c == '?' || c == '[') {
1042 INTOFF;
1043 expmeta(expdir, pattern, dstlist);
1044 INTON;
1045 break;
1046 }
1047 }
1048 if (dstlist->count == firstmatch) {
1049 /*
1050 * no matches
1051 */
1052 rmescapes(pattern);
1053 appendarglist(dstlist, pattern);
1054 } else {
1055 qsort(&dstlist->args[firstmatch],
1056 dstlist->count - firstmatch,
1057 sizeof(dstlist->args[0]), expsortcmp);
1058 }
1059 }
1060
1061
1062 /*
1063 * Do metacharacter (i.e. *, ?, [...]) expansion.
1064 */
1065
1066 static void
1067 expmeta(char *enddir, char *name, struct arglist *arglist)
1068 {
1069 const char *p;
1070 const char *q;
1071 const char *start;
1072 char *endname;
1073 int metaflag;
1074 struct stat statb;
1075 DIR *dirp;
1076 struct dirent *dp;
1077 int atend;
1078 int matchdot;
1079 int esc;
1080 int namlen;
1081
1082 metaflag = 0;
1083 start = name;
1084 for (p = name; esc = 0, *p; p += esc + 1) {
1085 if (*p == '*' || *p == '?')
1086 metaflag = 1;
1087 else if (*p == '[') {
1088 q = p + 1;
1089 if (*q == '!' || *q == '^')
1090 q++;
1091 for (;;) {
1092 if (*q == CTLESC)
1093 q++;
1094 if (*q == '/' || *q == '\0')
1095 break;
1096 if (*++q == ']') {
1097 metaflag = 1;
1098 break;
1099 }
1100 }
1101 } else if (*p == '\0')
1102 break;
1103 else {
1104 if (*p == CTLESC)
1105 esc++;
1106 if (p[esc] == '/') {
1107 if (metaflag)
1108 break;
1109 start = p + esc + 1;
1110 }
1111 }
1112 }
1113 if (metaflag == 0) { /* we've reached the end of the file name */
1114 if (enddir != expdir)
1115 metaflag++;
1116 for (p = name ; ; p++) {
1117 if (*p == CTLESC)
1118 p++;
1119 *enddir++ = *p;
1120 if (*p == '\0')
1121 break;
1122 if (enddir == expdir_end)
1123 return;
1124 }
1125 if (metaflag == 0 || lstat(expdir, &statb) >= 0)
1126 appendarglist(arglist, stsavestr(expdir));
1127 return;
1128 }
1129 endname = name + (p - name);
1130 if (start != name) {
1131 p = name;
1132 while (p < start) {
1133 if (*p == CTLESC)
1134 p++;
1135 *enddir++ = *p++;
1136 if (enddir == expdir_end)
1137 return;
1138 }
1139 }
1140 if (enddir == expdir) {
1141 p = ".";
1142 } else if (enddir == expdir + 1 && *expdir == '/') {
1143 p = "/";
1144 } else {
1145 p = expdir;
1146 enddir[-1] = '\0';
1147 }
1148 if ((dirp = opendir(p)) == NULL)
1149 return;
1150 if (enddir != expdir)
1151 enddir[-1] = '/';
1152 if (*endname == 0) {
1153 atend = 1;
1154 } else {
1155 atend = 0;
1156 *endname = '\0';
1157 endname += esc + 1;
1158 }
1159 matchdot = 0;
1160 p = start;
1161 if (*p == CTLESC)
1162 p++;
1163 if (*p == '.')
1164 matchdot++;
1165 while (! int_pending() && (dp = readdir(dirp)) != NULL) {
1166 if (dp->d_name[0] == '.' && ! matchdot)
1167 continue;
1168 if (patmatch(start, dp->d_name)) {
1169 namlen = dp->d_namlen;
1170 if (enddir + namlen + 1 > expdir_end)
1171 continue;
1172 memcpy(enddir, dp->d_name, namlen + 1);
1173 if (atend)
1174 appendarglist(arglist, stsavestr(expdir));
1175 else {
1176 if (dp->d_type != DT_UNKNOWN &&
1177 dp->d_type != DT_DIR &&
1178 dp->d_type != DT_LNK)
1179 continue;
1180 if (enddir + namlen + 2 > expdir_end)
1181 continue;
1182 enddir[namlen] = '/';
1183 enddir[namlen + 1] = '\0';
1184 expmeta(enddir + namlen + 1, endname, arglist);
1185 }
1186 }
1187 }
1188 closedir(dirp);
1189 if (! atend)
1190 endname[-esc - 1] = esc ? CTLESC : '/';
1191 }
1192
1193
1194 static int
1195 expsortcmp(const void *p1, const void *p2)
1196 {
1197 const char *s1 = *(const char * const *)p1;
1198 const char *s2 = *(const char * const *)p2;
1199
1200 return (strcoll(s1, s2));
1201 }
1202
1203
1204
1205 static wchar_t
1206 get_wc(const char **p)
1207 {
1208 wchar_t c;
1209 int chrlen;
1210
1211 chrlen = mbtowc(&c, *p, 4);
1212 if (chrlen == 0)
1213 return 0;
1214 else if (chrlen == -1)
1215 c = 0;
1216 else
1217 *p += chrlen;
1218 return c;
1219 }
1220
1221
1222 /*
1223 * See if a character matches a character class, starting at the first colon
1224 * of "[:class:]".
1225 * If a valid character class is recognized, a pointer to the next character
1226 * after the final closing bracket is stored into *end, otherwise a null
1227 * pointer is stored into *end.
1228 */
1229 static int
1230 match_charclass(const char *p, wchar_t chr, const char **end)
1231 {
1232 char name[20];
1233 const char *nameend;
1234 wctype_t cclass;
1235
1236 *end = NULL;
1237 p++;
1238 nameend = strstr(p, ":]");
1239 if (nameend == NULL || (size_t)(nameend - p) >= sizeof(name) ||
1240 nameend == p)
1241 return 0;
1242 memcpy(name, p, nameend - p);
1243 name[nameend - p] = '\0';
1244 *end = nameend + 2;
1245 cclass = wctype(name);
1246 /* An unknown class matches nothing but is valid nevertheless. */
1247 if (cclass == 0)
1248 return 0;
1249 return iswctype(chr, cclass);
1250 }
1251
1252
1253 /*
1254 * Returns true if the pattern matches the string.
1255 */
1256
1257 static int
1258 patmatch(const char *pattern, const char *string)
1259 {
1260 const char *p, *q, *end;
1261 const char *bt_p, *bt_q;
1262 char c;
1263 wchar_t wc, wc2;
1264
1265 p = pattern;
1266 q = string;
1267 bt_p = NULL;
1268 bt_q = NULL;
1269 for (;;) {
1270 switch (c = *p++) {
1271 case '\0':
1272 if (*q != '\0')
1273 goto backtrack;
1274 return 1;
1275 case CTLESC:
1276 if (*q++ != *p++)
1277 goto backtrack;
1278 break;
1279 case '?':
1280 if (*q == '\0')
1281 return 0;
1282 if (localeisutf8) {
1283 wc = get_wc(&q);
1284 /*
1285 * A '?' does not match invalid UTF-8 but a
1286 * '*' does, so backtrack.
1287 */
1288 if (wc == 0)
1289 goto backtrack;
1290 } else
1291 q++;
1292 break;
1293 case '*':
1294 c = *p;
1295 while (c == '*')
1296 c = *++p;
1297 /*
1298 * If the pattern ends here, we know the string
1299 * matches without needing to look at the rest of it.
1300 */
1301 if (c == '\0')
1302 return 1;
1303 /*
1304 * First try the shortest match for the '*' that
1305 * could work. We can forget any earlier '*' since
1306 * there is no way having it match more characters
1307 * can help us, given that we are already here.
1308 */
1309 bt_p = p;
1310 bt_q = q;
1311 break;
1312 case '[': {
1313 const char *savep, *saveq;
1314 int invert, found;
1315 wchar_t chr;
1316
1317 savep = p, saveq = q;
1318 invert = 0;
1319 if (*p == '!' || *p == '^') {
1320 invert++;
1321 p++;
1322 }
1323 found = 0;
1324 if (*q == '\0')
1325 return 0;
1326 if (localeisutf8) {
1327 chr = get_wc(&q);
1328 if (chr == 0)
1329 goto backtrack;
1330 } else
1331 chr = (unsigned char)*q++;
1332 c = *p++;
1333 do {
1334 if (c == '\0') {
1335 p = savep, q = saveq;
1336 c = '[';
1337 goto dft;
1338 }
1339 if (c == '[' && *p == ':') {
1340 found |= match_charclass(p, chr, &end);
1341 if (end != NULL)
1342 p = end;
1343 }
1344 if (c == CTLESC)
1345 c = *p++;
1346 if (localeisutf8 && c & 0x80) {
1347 p--;
1348 wc = get_wc(&p);
1349 if (wc == 0) /* bad utf-8 */
1350 return 0;
1351 } else
1352 wc = (unsigned char)c;
1353 if (*p == '-' && p[1] != ']') {
1354 p++;
1355 if (*p == CTLESC)
1356 p++;
1357 if (localeisutf8) {
1358 wc2 = get_wc(&p);
1359 if (wc2 == 0) /* bad utf-8 */
1360 return 0;
1361 } else
1362 wc2 = (unsigned char)*p++;
1363 if ( collate_range_cmp(chr, wc) >= 0
1364 && collate_range_cmp(chr, wc2) <= 0
1365 )
1366 found = 1;
1367 } else {
1368 if (chr == wc)
1369 found = 1;
1370 }
1371 } while ((c = *p++) != ']');
1372 if (found == invert)
1373 goto backtrack;
1374 break;
1375 }
1376 dft: default:
1377 if (*q == '\0')
1378 return 0;
1379 if (*q++ == c)
1380 break;
1381 backtrack:
1382 /*
1383 * If we have a mismatch (other than hitting the end
1384 * of the string), go back to the last '*' seen and
1385 * have it match one additional character.
1386 */
1387 if (bt_p == NULL)
1388 return 0;
1389 if (*bt_q == '\0')
1390 return 0;
1391 bt_q++;
1392 p = bt_p;
1393 q = bt_q;
1394 break;
1395 }
1396 }
1397 }
1398
1399
1400
1401 /*
1402 * Remove any CTLESC and CTLQUOTEMARK characters from a string.
1403 */
1404
1405 void
1406 rmescapes(char *str)
1407 {
1408 char *p, *q;
1409
1410 p = str;
1411 while (*p != CTLESC && *p != CTLQUOTEMARK && *p != CTLQUOTEEND) {
1412 if (*p++ == '\0')
1413 return;
1414 }
1415 q = p;
1416 while (*p) {
1417 if (*p == CTLQUOTEMARK || *p == CTLQUOTEEND) {
1418 p++;
1419 continue;
1420 }
1421 if (*p == CTLESC)
1422 p++;
1423 *q++ = *p++;
1424 }
1425 *q = '\0';
1426 }
1427
1428
1429
1430 /*
1431 * See if a pattern matches in a case statement.
1432 */
1433
1434 int
1435 casematch(union node *pattern, const char *val)
1436 {
1437 struct stackmark smark;
1438 int result;
1439 char *p;
1440
1441 setstackmark(&smark);
1442 argbackq = pattern->narg.backquote;
1443 STARTSTACKSTR(expdest);
1444 argstr(pattern->narg.text, EXP_TILDE | EXP_CASE, NULL);
1445 STPUTC('\0', expdest);
1446 p = grabstackstr(expdest);
1447 result = patmatch(p, val);
1448 popstackmark(&smark);
1449 return result;
1450 }
1451
1452 /*
1453 * Our own itoa().
1454 */
1455
1456 static void
1457 cvtnum(int num, char *buf)
1458 {
1459 char temp[32];
1460 int neg = num < 0;
1461 char *p = temp + 31;
1462
1463 temp[31] = '\0';
1464
1465 do {
1466 *--p = num % 10 + '0';
1467 } while ((num /= 10) != 0);
1468
1469 if (neg)
1470 *--p = '-';
1471
1472 memcpy(buf, p, temp + 32 - p);
1473 }
1474
1475 /*
1476 * Do most of the work for wordexp(3).
1477 */
1478
1479 int
1480 wordexpcmd(int argc, char **argv)
1481 {
1482 size_t len;
1483 int i;
1484
1485 out1fmt("%08x", argc - 1);
1486 for (i = 1, len = 0; i < argc; i++)
1487 len += strlen(argv[i]);
1488 out1fmt("%08x", (int)len);
1489 for (i = 1; i < argc; i++)
1490 outbin(argv[i], strlen(argv[i]) + 1, out1);
1491 return (0);
1492 }
1493
1494 /*
1495 * Do most of the work for wordexp(3), new version.
1496 */
1497
1498 int
1499 freebsd_wordexpcmd(int argc __unused, char **argv __unused)
1500 {
1501 struct arglist arglist;
1502 union node *args, *n;
1503 size_t len;
1504 int ch;
1505 int protected = 0;
1506 int fd = -1;
1507 int i;
1508
1509 while ((ch = nextopt("f:p")) != '\0') {
1510 switch (ch) {
1511 case 'f':
1512 fd = number(shoptarg);
1513 break;
1514 case 'p':
1515 protected = 1;
1516 break;
1517 }
1518 }
1519 if (*argptr != NULL)
1520 error("wrong number of arguments");
1521 if (fd < 0)
1522 error("missing fd");
1523 INTOFF;
1524 setinputfd(fd, 1);
1525 INTON;
1526 args = parsewordexp();
1527 popfile(); /* will also close fd */
1528 if (protected)
1529 for (n = args; n != NULL; n = n->narg.next) {
1530 if (n->narg.backquote != NULL) {
1531 outcslow('C', out1);
1532 error("command substitution disabled");
1533 }
1534 }
1535 outcslow(' ', out1);
1536 emptyarglist(&arglist);
1537 for (n = args; n != NULL; n = n->narg.next)
1538 expandarg(n, &arglist, EXP_FULL | EXP_TILDE);
1539 for (i = 0, len = 0; i < arglist.count; i++)
1540 len += strlen(arglist.args[i]);
1541 out1fmt("%016x %016zx", arglist.count, len);
1542 for (i = 0; i < arglist.count; i++)
1543 outbin(arglist.args[i], strlen(arglist.args[i]) + 1, out1);
1544 return (0);
1545 }