1 /*
2 * Copyright (c) 2013 Gary Mills
3 */
4 /* $OpenBSD: glob.c,v 1.39 2012/01/20 07:09:42 tedu Exp $ */
5 /*
6 * Copyright (c) 1989, 1993
7 * The Regents of the University of California. All rights reserved.
8 *
9 * This code is derived from software contributed to Berkeley by
10 * Guido van Rossum.
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 * 3. 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 /*
38 * glob(3) -- a superset of the one defined in POSIX 1003.2.
39 *
40 * The [!...] convention to negate a range is supported (SysV, Posix, ksh).
41 *
42 * Optional extra services, controlled by flags not defined by POSIX:
43 *
44 * GLOB_QUOTE:
45 * Escaping convention: \ inhibits any special meaning the following
46 * character might have (except \ at end of string is retained).
47 * GLOB_MAGCHAR:
48 * Set in gl_flags if pattern contained a globbing character.
49 * GLOB_NOMAGIC:
50 * Same as GLOB_NOCHECK, but it will only append pattern if it did
51 * not contain any magic characters. [Used in csh style globbing]
52 * GLOB_ALTDIRFUNC:
53 * Use alternately specified directory access functions.
54 * GLOB_TILDE:
55 * expand ~user/foo to the /home/dir/of/user/foo
56 * GLOB_BRACE:
57 * expand {1,2}{a,b} to 1a 1b 2a 2b
58 * gl_matchc:
59 * Number of matches in the current invocation of glob.
60 */
61
62 #include "lint.h"
63
64 #include <sys/param.h>
65 #include <sys/stat.h>
66
67 #include <ctype.h>
68 #include <dirent.h>
69 #include <errno.h>
70 #include <glob.h>
71 #include <limits.h>
72 #include <pwd.h>
73 #include <stdio.h>
74 #include <stdlib.h>
75 #include <string.h>
76 #include <unistd.h>
77 #include <wchar.h>
78 #include <wctype.h>
79
80 /*
81 * This is the legacy glob_t prior to illumos enhancement 1097,
82 * used when old programs call the old libc glob functions.
83 * (New programs call the _glob_ext, _globfree_ext functions.)
84 * This struct should be considered "carved in stone".
85 */
86 typedef struct old_glob {
87 size_t gl_pathc; /* Count of paths matched by pattern */
88 char **gl_pathv; /* List of matched pathnames */
89 size_t gl_offs; /* # of slots reserved in gl_pathv */
90 /* following are internal to the implementation */
91 char **gl_pathp; /* gl_pathv + gl_offs */
92 int gl_pathn; /* # of elements allocated */
93 } old_glob_t;
94
95 /*
96 * For old programs, the external names need to be the old names:
97 * glob() and globfree() . We've redefined those already to
98 * _glob_ext() and _globfree_ext() . Now redefine old_glob()
99 * and old_globfree() to glob() and globfree() .
100 */
101 #ifdef __PRAGMA_REDEFINE_EXTNAME
102 #pragma redefine_extname old_glob glob
103 #pragma redefine_extname old_globfree globfree
104 #endif /* __PRAGMA_REDEFINE_EXTNAME */
105 extern int old_glob(const char *, int, int (*)(const char *, int),
106 old_glob_t *);
107 extern void old_globfree(old_glob_t *);
108
109 #define DOLLAR '$'
110 #define DOT '.'
111 #define EOS '\0'
112 #define LBRACKET '['
113 #define NOT '!'
114 #define QUESTION '?'
115 #define QUOTE '\\'
116 #define RANGE '-'
117 #define RBRACKET ']'
118 #define SEP '/'
119 #define STAR '*'
120 #define TILDE '~'
121 #define UNDERSCORE '_'
122 #define LBRACE '{'
123 #define RBRACE '}'
124 #define SLASH '/'
125 #define COMMA ','
126 #define COLON ':'
127
128 #define M_QUOTE 0x800000
129 #define M_PROTECT 0x400000
130
131 typedef struct wcat {
132 wchar_t w_wc;
133 uint_t w_at;
134 } wcat_t;
135
136 #define M_ALL '*' /* Plus M_QUOTE */
137 #define M_END ']' /* Plus M_QUOTE */
138 #define M_NOT '!' /* Plus M_QUOTE */
139 #define M_ONE '?' /* Plus M_QUOTE */
140 #define M_RNG '-' /* Plus M_QUOTE */
141 #define M_SET '[' /* Plus M_QUOTE */
142 #define M_CLASS ':' /* Plus M_QUOTE */
143 #define ismeta(c) (((c).w_at&M_QUOTE) != 0)
144
145 #define INITIAL 8 /* initial pathv allocation */
146
147 #define GLOB_LIMIT_MALLOC 65536
148 #define GLOB_LIMIT_STAT 2048
149 #define GLOB_LIMIT_READDIR 16384
150
151 /* Limit of recursion during matching attempts. */
152 #define GLOB_LIMIT_RECUR 64
153
154 struct glob_lim {
155 size_t glim_malloc;
156 size_t glim_stat;
157 size_t glim_readdir;
158 };
159
160 struct glob_path_stat {
161 char *gps_path;
162 struct stat *gps_stat;
163 };
164
165 static int compare(const void *, const void *);
166 static int compare_gps(const void *, const void *);
167 static int g_Ctoc(const wcat_t *, char *, uint_t);
168 static int g_lstat(wcat_t *, struct stat *, glob_t *);
169 static DIR *g_opendir(wcat_t *, glob_t *);
170 static wcat_t *g_strchr(const wcat_t *, wchar_t);
171 static int g_stat(wcat_t *, struct stat *, glob_t *);
172 static int glob0(const wcat_t *, glob_t *, struct glob_lim *,
173 int (*)(const char *, int));
174 static int glob1(wcat_t *, wcat_t *, glob_t *, struct glob_lim *,
175 int (*)(const char *, int));
176 static int glob2(wcat_t *, wcat_t *, wcat_t *, wcat_t *, wcat_t *,
177 wcat_t *, glob_t *, struct glob_lim *,
178 int (*)(const char *, int));
179 static int glob3(wcat_t *, wcat_t *, wcat_t *, wcat_t *, wcat_t *,
180 wcat_t *, wcat_t *, glob_t *, struct glob_lim *,
181 int (*)(const char *, int));
182 static int globextend(const wcat_t *, glob_t *, struct glob_lim *,
183 struct stat *);
184 static
185 const wcat_t *globtilde(const wcat_t *, wcat_t *, size_t, glob_t *);
186 static int globexp1(const wcat_t *, glob_t *, struct glob_lim *,
187 int (*)(const char *, int));
188 static int globexp2(const wcat_t *, const wcat_t *, glob_t *,
189 struct glob_lim *, int (*)(const char *, int));
190 static int match(wcat_t *, wcat_t *, wcat_t *, int);
191
192 /*
193 * Extended glob() function, selected by #pragma redefine_extname
194 * in glob.h with the external name _glob_ext() .
195 */
196 int
197 _glob_ext(const char *pattern, int flags, int (*errfunc)(const char *, int),
198 glob_t *pglob)
199 {
200 const char *patnext;
201 int n;
202 size_t patlen;
203 wchar_t c;
204 wcat_t *bufnext, *bufend, patbuf[MAXPATHLEN];
205 struct glob_lim limit = { 0, 0, 0 };
206
207 if ((patlen = strnlen(pattern, PATH_MAX)) == PATH_MAX)
208 return (GLOB_NOMATCH);
209
210 patnext = pattern;
211 if (!(flags & GLOB_APPEND)) {
212 pglob->gl_pathc = 0;
213 pglob->gl_pathn = 0;
214 pglob->gl_pathv = NULL;
215 if ((flags & GLOB_KEEPSTAT) != 0)
216 pglob->gl_statv = NULL;
217 if (!(flags & GLOB_DOOFFS))
218 pglob->gl_offs = 0;
219 }
220 pglob->gl_flags = flags & ~GLOB_MAGCHAR;
221 pglob->gl_matchc = 0;
222
223 if (pglob->gl_offs >= INT_MAX || pglob->gl_pathc >= INT_MAX ||
224 pglob->gl_pathc >= INT_MAX - pglob->gl_offs - 1)
225 return (GLOB_NOSPACE);
226
227 bufnext = patbuf;
228 bufend = bufnext + MAXPATHLEN - 1;
229 patlen += 1;
230 if (flags & GLOB_NOESCAPE) {
231 while (bufnext < bufend) {
232 if ((n = mbtowc(&c, patnext, patlen)) > 0) {
233 patnext += n;
234 patlen -= n;
235 bufnext->w_at = 0;
236 (bufnext++)->w_wc = c;
237 } else if (n == 0) {
238 break;
239 } else {
240 return (GLOB_NOMATCH);
241 }
242 }
243 } else {
244 /* Protect the quoted characters. */
245 while (bufnext < bufend) {
246 if ((n = mbtowc(&c, patnext, patlen)) > 0) {
247 patnext += n;
248 patlen -= n;
249 if (c == QUOTE) {
250 n = mbtowc(&c, patnext, patlen);
251 if (n < 0)
252 return (GLOB_NOMATCH);
253 if (n > 0) {
254 patnext += n;
255 patlen -= n;
256 }
257 if (n == 0)
258 c = QUOTE;
259 bufnext->w_at = M_PROTECT;
260 (bufnext++)->w_wc = c;
261 } else {
262 bufnext->w_at = 0;
263 (bufnext++)->w_wc = c;
264 }
265 } else if (n == 0) {
266 break;
267 } else {
268 return (GLOB_NOMATCH);
269 }
270 }
271 }
272 bufnext->w_at = 0;
273 bufnext->w_wc = EOS;
274
275 if (flags & GLOB_BRACE)
276 return (globexp1(patbuf, pglob, &limit, errfunc));
277 else
278 return (glob0(patbuf, pglob, &limit, errfunc));
279 }
280
281 /*
282 * Expand recursively a glob {} pattern. When there is no more expansion
283 * invoke the standard globbing routine to glob the rest of the magic
284 * characters
285 */
286 static int
287 globexp1(const wcat_t *pattern, glob_t *pglob, struct glob_lim *limitp,
288 int (*errfunc)(const char *, int))
289 {
290 const wcat_t *ptr = pattern;
291
292 /* Protect a single {}, for find(1), like csh */
293 if (pattern[0].w_wc == LBRACE && pattern[1].w_wc == RBRACE &&
294 pattern[2].w_wc == EOS)
295 return (glob0(pattern, pglob, limitp, errfunc));
296
297 if ((ptr = (const wcat_t *) g_strchr(ptr, LBRACE)) != NULL)
298 return (globexp2(ptr, pattern, pglob, limitp, errfunc));
299
300 return (glob0(pattern, pglob, limitp, errfunc));
301 }
302
303
304 /*
305 * Recursive brace globbing helper. Tries to expand a single brace.
306 * If it succeeds then it invokes globexp1 with the new pattern.
307 * If it fails then it tries to glob the rest of the pattern and returns.
308 */
309 static int
310 globexp2(const wcat_t *ptr, const wcat_t *pattern, glob_t *pglob,
311 struct glob_lim *limitp, int (*errfunc)(const char *, int))
312 {
313 int i, rv;
314 wcat_t *lm, *ls;
315 const wcat_t *pe, *pm, *pl;
316 wcat_t patbuf[MAXPATHLEN];
317
318 /* copy part up to the brace */
319 for (lm = patbuf, pm = pattern; pm != ptr; *lm++ = *pm++)
320 ;
321 lm->w_at = 0;
322 lm->w_wc = EOS;
323 ls = lm;
324
325 /* Find the balanced brace */
326 for (i = 0, pe = ++ptr; pe->w_wc != EOS; pe++)
327 if (pe->w_wc == LBRACKET) {
328 /* Ignore everything between [] */
329 for (pm = pe++; pe->w_wc != RBRACKET &&
330 pe->w_wc != EOS; pe++)
331 ;
332 if (pe->w_wc == EOS) {
333 /*
334 * We could not find a matching RBRACKET.
335 * Ignore and just look for RBRACE
336 */
337 pe = pm;
338 }
339 } else if (pe->w_wc == LBRACE) {
340 i++;
341 } else if (pe->w_wc == RBRACE) {
342 if (i == 0)
343 break;
344 i--;
345 }
346
347 /* Non matching braces; just glob the pattern */
348 if (i != 0 || pe->w_wc == EOS)
349 return (glob0(patbuf, pglob, limitp, errfunc));
350
351 for (i = 0, pl = pm = ptr; pm <= pe; pm++) {
352 switch (pm->w_wc) {
353 case LBRACKET:
354 /* Ignore everything between [] */
355 for (pl = pm++; pm->w_wc != RBRACKET && pm->w_wc != EOS;
356 pm++)
357 ;
358 if (pm->w_wc == EOS) {
359 /*
360 * We could not find a matching RBRACKET.
361 * Ignore and just look for RBRACE
362 */
363 pm = pl;
364 }
365 break;
366
367 case LBRACE:
368 i++;
369 break;
370
371 case RBRACE:
372 if (i) {
373 i--;
374 break;
375 }
376 /* FALLTHROUGH */
377 case COMMA:
378 if (i && pm->w_wc == COMMA)
379 break;
380 else {
381 /* Append the current string */
382 for (lm = ls; (pl < pm); *lm++ = *pl++)
383 ;
384
385 /*
386 * Append the rest of the pattern after the
387 * closing brace
388 */
389 for (pl = pe + 1;
390 (*lm++ = *pl++).w_wc != EOS; /* */)
391 ;
392
393 /* Expand the current pattern */
394 rv = globexp1(patbuf, pglob, limitp, errfunc);
395 if (rv && rv != GLOB_NOMATCH)
396 return (rv);
397
398 /* move after the comma, to the next string */
399 pl = pm + 1;
400 }
401 break;
402
403 default:
404 break;
405 }
406 }
407 return (0);
408 }
409
410
411
412 /*
413 * expand tilde from the passwd file.
414 */
415 static const wcat_t *
416 globtilde(const wcat_t *pattern, wcat_t *patbuf, size_t patbuf_len,
417 glob_t *pglob)
418 {
419 struct passwd *pwd;
420 char *h;
421 const wcat_t *p;
422 wcat_t *b, *eb, *q;
423 int n;
424 size_t lenh;
425 wchar_t c;
426
427 if (pattern->w_wc != TILDE || !(pglob->gl_flags & GLOB_TILDE))
428 return (pattern);
429
430 /* Copy up to the end of the string or / */
431 eb = &patbuf[patbuf_len - 1];
432 for (p = pattern + 1, q = patbuf;
433 q < eb && p->w_wc != EOS && p->w_wc != SLASH; *q++ = *p++)
434 ;
435
436 q->w_at = 0;
437 q->w_wc = EOS;
438
439 /* What to do if patbuf is full? */
440
441 if (patbuf[0].w_wc == EOS) {
442 /*
443 * handle a plain ~ or ~/ by expanding $HOME
444 * first and then trying the password file
445 */
446 if (issetugid() != 0)
447 return (pattern);
448 if ((h = getenv("HOME")) == NULL) {
449 if ((pwd = getpwuid(getuid())) == NULL)
450 return (pattern);
451 else
452 h = pwd->pw_dir;
453 }
454 } else {
455 /*
456 * Expand a ~user
457 */
458 if ((pwd = getpwnam((char *)patbuf)) == NULL)
459 return (pattern);
460 else
461 h = pwd->pw_dir;
462 }
463
464 /* Copy the home directory */
465 lenh = strlen(h) + 1;
466 for (b = patbuf; b < eb && *h != EOS; b++) {
467 if ((n = mbtowc(&c, h, lenh)) > 0) {
468 h += n;
469 lenh -= n;
470 b->w_at = 0;
471 b->w_wc = c;
472 } else if (n < 0) {
473 return (pattern);
474 } else {
475 break;
476 }
477 }
478
479 /* Append the rest of the pattern */
480 while (b < eb && (*b++ = *p++).w_wc != EOS)
481 ;
482 b->w_at = 0;
483 b->w_wc = EOS;
484
485 return (patbuf);
486 }
487
488 static int
489 g_charclass(const wcat_t **patternp, wcat_t **bufnextp)
490 {
491 const wcat_t *pattern = *patternp + 1;
492 wcat_t *bufnext = *bufnextp;
493 const wcat_t *colon;
494 char cbuf[MB_LEN_MAX + 32];
495 wctype_t cc;
496 size_t len;
497
498 if ((colon = g_strchr(pattern, COLON)) == NULL ||
499 colon[1].w_wc != RBRACKET)
500 return (1); /* not a character class */
501
502 len = (size_t)(colon - pattern);
503 if (len + MB_LEN_MAX + 1 > sizeof (cbuf))
504 return (-1); /* invalid character class */
505 {
506 wchar_t w;
507 const wcat_t *s1 = pattern;
508 char *s2 = cbuf;
509 size_t n = len;
510
511 /* Copy the string. */
512 while (n > 0) {
513 w = (s1++)->w_wc;
514 /* Character class names must be ASCII. */
515 if (iswascii(w)) {
516 n--;
517 *s2++ = w;
518 } else {
519 return (-1); /* invalid character class */
520 }
521 }
522 *s2 = EOS;
523 }
524 if ((cc = wctype(cbuf)) == 0)
525 return (-1); /* invalid character class */
526 bufnext->w_at = M_QUOTE;
527 (bufnext++)->w_wc = M_CLASS;
528 bufnext->w_at = 0;
529 (bufnext++)->w_wc = cc;
530 *bufnextp = bufnext;
531 *patternp += len + 3;
532
533 return (0);
534 }
535
536 /*
537 * The main glob() routine: compiles the pattern (optionally processing
538 * quotes), calls glob1() to do the real pattern matching, and finally
539 * sorts the list (unless unsorted operation is requested). Returns 0
540 * if things went well, nonzero if errors occurred. It is not an error
541 * to find no matches.
542 */
543 static int
544 glob0(const wcat_t *pattern, glob_t *pglob, struct glob_lim *limitp,
545 int (*errfunc)(const char *, int))
546 {
547 const wcat_t *qpatnext;
548 int err, oldpathc;
549 wchar_t c;
550 int a;
551 wcat_t *bufnext, patbuf[MAXPATHLEN];
552
553 qpatnext = globtilde(pattern, patbuf, MAXPATHLEN, pglob);
554 oldpathc = pglob->gl_pathc;
555 bufnext = patbuf;
556
557 /*
558 * We don't need to check for buffer overflow any more.
559 * The pattern has already been copied to an internal buffer.
560 */
561 while ((a = qpatnext->w_at), (c = (qpatnext++)->w_wc) != EOS) {
562 switch (c) {
563 case LBRACKET:
564 if (a != 0) {
565 bufnext->w_at = a;
566 (bufnext++)->w_wc = c;
567 break;
568 }
569 a = qpatnext->w_at;
570 c = qpatnext->w_wc;
571 if (a == 0 && c == NOT)
572 ++qpatnext;
573 if (qpatnext->w_wc == EOS ||
574 g_strchr(qpatnext+1, RBRACKET) == NULL) {
575 bufnext->w_at = 0;
576 (bufnext++)->w_wc = LBRACKET;
577 if (a == 0 && c == NOT)
578 --qpatnext;
579 break;
580 }
581 bufnext->w_at = M_QUOTE;
582 (bufnext++)->w_wc = M_SET;
583 if (a == 0 && c == NOT) {
584 bufnext->w_at = M_QUOTE;
585 (bufnext++)->w_wc = M_NOT;
586 }
587 a = qpatnext->w_at;
588 c = (qpatnext++)->w_wc;
589 do {
590 if (a == 0 && c == LBRACKET &&
591 qpatnext->w_wc == COLON) {
592 do {
593 err = g_charclass(&qpatnext,
594 &bufnext);
595 if (err)
596 break;
597 a = qpatnext->w_at;
598 c = (qpatnext++)->w_wc;
599 } while (a == 0 && c == LBRACKET &&
600 qpatnext->w_wc == COLON);
601 if (err == -1 &&
602 !(pglob->gl_flags & GLOB_NOCHECK))
603 return (GLOB_NOMATCH);
604 if (a == 0 && c == RBRACKET)
605 break;
606 }
607 bufnext->w_at = a;
608 (bufnext++)->w_wc = c;
609 if (qpatnext->w_at == 0 &&
610 qpatnext->w_wc == RANGE) {
611 a = qpatnext[1].w_at;
612 c = qpatnext[1].w_wc;
613 if (qpatnext[1].w_at != 0 ||
614 qpatnext[1].w_wc != RBRACKET) {
615 bufnext->w_at = M_QUOTE;
616 (bufnext++)->w_wc = M_RNG;
617 bufnext->w_at = a;
618 (bufnext++)->w_wc = c;
619 qpatnext += 2;
620 }
621 }
622 a = qpatnext->w_at;
623 c = (qpatnext++)->w_wc;
624 } while (a != 0 || c != RBRACKET);
625 pglob->gl_flags |= GLOB_MAGCHAR;
626 bufnext->w_at = M_QUOTE;
627 (bufnext++)->w_wc = M_END;
628 break;
629 case QUESTION:
630 if (a != 0) {
631 bufnext->w_at = a;
632 (bufnext++)->w_wc = c;
633 break;
634 }
635 pglob->gl_flags |= GLOB_MAGCHAR;
636 bufnext->w_at = M_QUOTE;
637 (bufnext++)->w_wc = M_ONE;
638 break;
639 case STAR:
640 if (a != 0) {
641 bufnext->w_at = a;
642 (bufnext++)->w_wc = c;
643 break;
644 }
645 pglob->gl_flags |= GLOB_MAGCHAR;
646 /*
647 * collapse adjacent stars to one,
648 * to avoid exponential behavior
649 */
650 if (bufnext == patbuf ||
651 bufnext[-1].w_at != M_QUOTE ||
652 bufnext[-1].w_wc != M_ALL) {
653 bufnext->w_at = M_QUOTE;
654 (bufnext++)->w_wc = M_ALL;
655 }
656 break;
657 default:
658 bufnext->w_at = a;
659 (bufnext++)->w_wc = c;
660 break;
661 }
662 }
663 bufnext->w_at = 0;
664 bufnext->w_wc = EOS;
665
666 if ((err = glob1(patbuf, patbuf+MAXPATHLEN-1, pglob, limitp, errfunc))
667 != 0)
668 return (err);
669
670 /*
671 * If there was no match we are going to append the pattern
672 * if GLOB_NOCHECK was specified or if GLOB_NOMAGIC was specified
673 * and the pattern did not contain any magic characters
674 * GLOB_NOMAGIC is there just for compatibility with csh.
675 */
676 if (pglob->gl_pathc == oldpathc) {
677 if ((pglob->gl_flags & GLOB_NOCHECK) ||
678 ((pglob->gl_flags & GLOB_NOMAGIC) &&
679 !(pglob->gl_flags & GLOB_MAGCHAR)))
680 return (globextend(pattern, pglob, limitp, NULL));
681 else
682 return (GLOB_NOMATCH);
683 }
684 if (!(pglob->gl_flags & GLOB_NOSORT)) {
685 if ((pglob->gl_flags & GLOB_KEEPSTAT)) {
686 /* Keep the paths and stat info synced during sort */
687 struct glob_path_stat *path_stat;
688 int i;
689 int n = pglob->gl_pathc - oldpathc;
690 int o = pglob->gl_offs + oldpathc;
691
692 if ((path_stat = calloc(n, sizeof (*path_stat))) ==
693 NULL)
694 return (GLOB_NOSPACE);
695 for (i = 0; i < n; i++) {
696 path_stat[i].gps_path = pglob->gl_pathv[o + i];
697 path_stat[i].gps_stat = pglob->gl_statv[o + i];
698 }
699 qsort(path_stat, n, sizeof (*path_stat), compare_gps);
700 for (i = 0; i < n; i++) {
701 pglob->gl_pathv[o + i] = path_stat[i].gps_path;
702 pglob->gl_statv[o + i] = path_stat[i].gps_stat;
703 }
704 free(path_stat);
705 } else {
706 qsort(pglob->gl_pathv + pglob->gl_offs + oldpathc,
707 pglob->gl_pathc - oldpathc, sizeof (char *),
708 compare);
709 }
710 }
711 return (0);
712 }
713
714 static int
715 compare(const void *p, const void *q)
716 {
717 return (strcmp(*(char **)p, *(char **)q));
718 }
719
720 static int
721 compare_gps(const void *_p, const void *_q)
722 {
723 const struct glob_path_stat *p = (const struct glob_path_stat *)_p;
724 const struct glob_path_stat *q = (const struct glob_path_stat *)_q;
725
726 return (strcmp(p->gps_path, q->gps_path));
727 }
728
729 static int
730 glob1(wcat_t *pattern, wcat_t *pattern_last, glob_t *pglob,
731 struct glob_lim *limitp, int (*errfunc)(const char *, int))
732 {
733 wcat_t pathbuf[MAXPATHLEN];
734
735 /* A null pathname is invalid -- POSIX 1003.1 sect. 2.4. */
736 if (pattern->w_wc == EOS)
737 return (0);
738 return (glob2(pathbuf, pathbuf+MAXPATHLEN-1,
739 pathbuf, pathbuf+MAXPATHLEN-1,
740 pattern, pattern_last, pglob, limitp, errfunc));
741 }
742
743 /*
744 * The functions glob2 and glob3 are mutually recursive; there is one level
745 * of recursion for each segment in the pattern that contains one or more
746 * meta characters.
747 */
748 static int
749 glob2(wcat_t *pathbuf, wcat_t *pathbuf_last, wcat_t *pathend,
750 wcat_t *pathend_last, wcat_t *pattern, wcat_t *pattern_last,
751 glob_t *pglob, struct glob_lim *limitp, int (*errfunc)(const char *, int))
752 {
753 struct stat sb;
754 wcat_t *p, *q;
755 int anymeta;
756
757 /*
758 * Loop over pattern segments until end of pattern or until
759 * segment with meta character found.
760 */
761 for (anymeta = 0; ; ) {
762 if (pattern->w_wc == EOS) { /* End of pattern? */
763 pathend->w_at = 0;
764 pathend->w_wc = EOS;
765
766 if ((pglob->gl_flags & GLOB_LIMIT) &&
767 limitp->glim_stat++ >= GLOB_LIMIT_STAT) {
768 errno = 0;
769 pathend->w_at = 0;
770 (pathend++)->w_wc = SEP;
771 pathend->w_at = 0;
772 pathend->w_wc = EOS;
773 return (GLOB_NOSPACE);
774 }
775 if (g_lstat(pathbuf, &sb, pglob))
776 return (0);
777
778 if (((pglob->gl_flags & GLOB_MARK) &&
779 (pathend[-1].w_at != 0 ||
780 pathend[-1].w_wc != SEP)) &&
781 (S_ISDIR(sb.st_mode) ||
782 (S_ISLNK(sb.st_mode) &&
783 (g_stat(pathbuf, &sb, pglob) == 0) &&
784 S_ISDIR(sb.st_mode)))) {
785 if (pathend+1 > pathend_last)
786 return (GLOB_NOSPACE);
787 pathend->w_at = 0;
788 (pathend++)->w_wc = SEP;
789 pathend->w_at = 0;
790 pathend->w_wc = EOS;
791 }
792 ++pglob->gl_matchc;
793 return (globextend(pathbuf, pglob, limitp, &sb));
794 }
795
796 /* Find end of next segment, copy tentatively to pathend. */
797 q = pathend;
798 p = pattern;
799 while (p->w_wc != EOS && p->w_wc != SEP) {
800 if (ismeta(*p))
801 anymeta = 1;
802 if (q+1 > pathend_last)
803 return (GLOB_NOSPACE);
804 *q++ = *p++;
805 }
806
807 if (!anymeta) { /* No expansion, do next segment. */
808 pathend = q;
809 pattern = p;
810 while (pattern->w_wc == SEP) {
811 if (pathend+1 > pathend_last)
812 return (GLOB_NOSPACE);
813 *pathend++ = *pattern++;
814 }
815 } else {
816 /* Need expansion, recurse. */
817 return (glob3(pathbuf, pathbuf_last, pathend,
818 pathend_last, pattern, p, pattern_last,
819 pglob, limitp, errfunc));
820 }
821 }
822 /* NOTREACHED */
823 }
824
825 static int
826 glob3(wcat_t *pathbuf, wcat_t *pathbuf_last, wcat_t *pathend,
827 wcat_t *pathend_last, wcat_t *pattern, wcat_t *restpattern,
828 wcat_t *restpattern_last, glob_t *pglob, struct glob_lim *limitp,
829 int (*errfunc)(const char *, int))
830 {
831 struct dirent *dp;
832 DIR *dirp;
833 int err;
834 char buf[MAXPATHLEN];
835
836 /*
837 * The readdirfunc declaration can't be prototyped, because it is
838 * assigned, below, to two functions which are prototyped in glob.h
839 * and dirent.h as taking pointers to differently typed opaque
840 * structures.
841 */
842 struct dirent *(*readdirfunc)(void *);
843
844 if (pathend > pathend_last)
845 return (GLOB_NOSPACE);
846 pathend->w_at = 0;
847 pathend->w_wc = EOS;
848 errno = 0;
849
850 if ((dirp = g_opendir(pathbuf, pglob)) == NULL) {
851 /* TODO: don't call for ENOENT or ENOTDIR? */
852 if (errfunc) {
853 if (g_Ctoc(pathbuf, buf, sizeof (buf)))
854 return (GLOB_ABORTED);
855 if (errfunc(buf, errno) ||
856 pglob->gl_flags & GLOB_ERR)
857 return (GLOB_ABORTED);
858 }
859 return (0);
860 }
861
862 err = 0;
863
864 /* Search directory for matching names. */
865 if (pglob->gl_flags & GLOB_ALTDIRFUNC)
866 readdirfunc = pglob->gl_readdir;
867 else
868 readdirfunc = (struct dirent *(*)(void *))readdir;
869 while ((dp = (*readdirfunc)(dirp))) {
870 char *sc;
871 wcat_t *dc;
872 int n;
873 int lensc;
874 wchar_t w;
875
876 if ((pglob->gl_flags & GLOB_LIMIT) &&
877 limitp->glim_readdir++ >= GLOB_LIMIT_READDIR) {
878 errno = 0;
879 pathend->w_at = 0;
880 (pathend++)->w_wc = SEP;
881 pathend->w_at = 0;
882 pathend->w_wc = EOS;
883 err = GLOB_NOSPACE;
884 break;
885 }
886
887 /* Initial DOT must be matched literally. */
888 if (dp->d_name[0] == DOT && pattern->w_wc != DOT)
889 continue;
890 dc = pathend;
891 sc = dp->d_name;
892 lensc = strlen(sc) + 1;
893 while (dc < pathend_last) {
894 if ((n = mbtowc(&w, sc, lensc)) <= 0) {
895 sc += 1;
896 lensc -= 1;
897 dc->w_at = 0;
898 dc->w_wc = EOS;
899 } else {
900 sc += n;
901 lensc -= n;
902 dc->w_at = 0;
903 dc->w_wc = w;
904 }
905 dc++;
906 if (n <= 0)
907 break;
908 }
909 if (dc >= pathend_last) {
910 dc->w_at = 0;
911 dc->w_wc = EOS;
912 err = GLOB_NOSPACE;
913 break;
914 }
915 if (n < 0) {
916 err = GLOB_NOMATCH;
917 break;
918 }
919
920 if (!match(pathend, pattern, restpattern, GLOB_LIMIT_RECUR)) {
921 pathend->w_at = 0;
922 pathend->w_wc = EOS;
923 continue;
924 }
925 err = glob2(pathbuf, pathbuf_last, --dc, pathend_last,
926 restpattern, restpattern_last, pglob, limitp,
927 errfunc);
928 if (err)
929 break;
930 }
931
932 if (pglob->gl_flags & GLOB_ALTDIRFUNC)
933 (*pglob->gl_closedir)(dirp);
934 else
935 (void) closedir(dirp);
936 return (err);
937 }
938
939
940 /*
941 * Extend the gl_pathv member of a glob_t structure to accommodate a new item,
942 * add the new item, and update gl_pathc. Avoids excessive reallocation
943 * by doubling the number of elements each time. Uses gl_pathn to contain
944 * the number.
945 *
946 * Return 0 if new item added, error code if memory couldn't be allocated.
947 *
948 * Invariant of the glob_t structure:
949 * Either gl_pathc is zero and gl_pathv is NULL; or gl_pathc > 0 and
950 * gl_pathv points to (gl_offs + gl_pathc + 1) items.
951 */
952 static int
953 globextend(const wcat_t *path, glob_t *pglob, struct glob_lim *limitp,
954 struct stat *sb)
955 {
956 char **pathv;
957 ssize_t i;
958 size_t allocn, newn, len;
959 char *copy = NULL;
960 const wcat_t *p;
961 struct stat **statv;
962 char junk[MB_LEN_MAX];
963 int n;
964
965 allocn = pglob->gl_pathn;
966 newn = 2 + pglob->gl_pathc + pglob->gl_offs;
967
968 if (newn <= allocn) {
969 pathv = pglob->gl_pathv;
970 if ((pglob->gl_flags & GLOB_KEEPSTAT) != 0)
971 statv = pglob->gl_statv;
972 } else {
973 if (allocn == 0)
974 allocn = pglob->gl_offs + INITIAL;
975 allocn *= 2;
976 if (pglob->gl_offs >= INT_MAX ||
977 pglob->gl_pathc >= INT_MAX ||
978 allocn >= INT_MAX ||
979 SIZE_MAX / sizeof (*pathv) <= allocn ||
980 SIZE_MAX / sizeof (*statv) <= allocn) {
981 nospace:
982 for (i = pglob->gl_offs; i < (ssize_t)(newn - 2);
983 i++) {
984 if (pglob->gl_pathv && pglob->gl_pathv[i])
985 free(pglob->gl_pathv[i]);
986 if ((pglob->gl_flags & GLOB_KEEPSTAT) != 0 &&
987 pglob->gl_statv && pglob->gl_statv[i])
988 free(pglob->gl_statv[i]);
989 }
990 if (pglob->gl_pathv) {
991 free(pglob->gl_pathv);
992 pglob->gl_pathv = NULL;
993 }
994 if ((pglob->gl_flags & GLOB_KEEPSTAT) != 0 &&
995 pglob->gl_statv) {
996 free(pglob->gl_statv);
997 pglob->gl_statv = NULL;
998 }
999 return (GLOB_NOSPACE);
1000 }
1001 limitp->glim_malloc += allocn * sizeof (*pathv);
1002 pathv = realloc(pglob->gl_pathv, allocn * sizeof (*pathv));
1003 if (pathv == NULL)
1004 goto nospace;
1005 if ((pglob->gl_flags & GLOB_KEEPSTAT) != 0) {
1006 limitp->glim_malloc += allocn * sizeof (*statv);
1007 statv = realloc(pglob->gl_statv,
1008 allocn * sizeof (*statv));
1009 if (statv == NULL)
1010 goto nospace;
1011 }
1012 }
1013 pglob->gl_pathn = allocn;
1014
1015 if (pglob->gl_pathv == NULL && pglob->gl_offs > 0) {
1016 /* first time around -- clear initial gl_offs items */
1017 pathv += pglob->gl_offs;
1018 for (i = pglob->gl_offs; --i >= 0; )
1019 *--pathv = NULL;
1020 }
1021 pglob->gl_pathv = pathv;
1022
1023 if ((pglob->gl_flags & GLOB_KEEPSTAT) != 0) {
1024 if (pglob->gl_statv == NULL && pglob->gl_offs > 0) {
1025 /* first time around -- clear initial gl_offs items */
1026 statv += pglob->gl_offs;
1027 for (i = pglob->gl_offs; --i >= 0; )
1028 *--statv = NULL;
1029 }
1030 pglob->gl_statv = statv;
1031 if (sb == NULL)
1032 statv[pglob->gl_offs + pglob->gl_pathc] = NULL;
1033 else {
1034 limitp->glim_malloc += sizeof (**statv);
1035 if ((statv[pglob->gl_offs + pglob->gl_pathc] =
1036 malloc(sizeof (**statv))) == NULL)
1037 goto copy_error;
1038 (void) memcpy(statv[pglob->gl_offs + pglob->gl_pathc],
1039 sb, sizeof (*sb));
1040 }
1041 statv[pglob->gl_offs + pglob->gl_pathc + 1] = NULL;
1042 }
1043
1044 len = MB_LEN_MAX;
1045 p = path;
1046 while ((n = wctomb(junk, p->w_wc)) > 0) {
1047 len += n;
1048 if ((p++)->w_wc == EOS)
1049 break;
1050 }
1051 if (n < 0)
1052 return (GLOB_NOMATCH);
1053
1054 limitp->glim_malloc += len;
1055 if ((copy = malloc(len)) != NULL) {
1056 if (g_Ctoc(path, copy, len)) {
1057 free(copy);
1058 return (GLOB_NOSPACE);
1059 }
1060 pathv[pglob->gl_offs + pglob->gl_pathc++] = copy;
1061 }
1062 pathv[pglob->gl_offs + pglob->gl_pathc] = NULL;
1063
1064 if ((pglob->gl_flags & GLOB_LIMIT) &&
1065 limitp->glim_malloc >= GLOB_LIMIT_MALLOC) {
1066 errno = 0;
1067 return (GLOB_NOSPACE);
1068 }
1069 copy_error:
1070 return (copy == NULL ? GLOB_NOSPACE : 0);
1071 }
1072
1073
1074 /*
1075 * pattern matching function for filenames. Each occurrence of the *
1076 * pattern causes a recursion level.
1077 */
1078 static int
1079 match(wcat_t *name, wcat_t *pat, wcat_t *patend, int recur)
1080 {
1081 int ok, negate_range;
1082 wcat_t c, k;
1083
1084 if (recur-- == 0)
1085 return (1);
1086
1087 while (pat < patend) {
1088 c = *pat++;
1089 switch (c.w_wc) {
1090 case M_ALL:
1091 if (c.w_at != M_QUOTE) {
1092 k = *name++;
1093 if (k.w_at != c.w_at || k.w_wc != c.w_wc)
1094 return (0);
1095 break;
1096 }
1097 while (pat < patend && pat->w_at == M_QUOTE &&
1098 pat->w_wc == M_ALL)
1099 pat++; /* eat consecutive '*' */
1100 if (pat == patend)
1101 return (1);
1102 do {
1103 if (match(name, pat, patend, recur))
1104 return (1);
1105 } while ((name++)->w_wc != EOS);
1106 return (0);
1107 case M_ONE:
1108 if (c.w_at != M_QUOTE) {
1109 k = *name++;
1110 if (k.w_at != c.w_at || k.w_wc != c.w_wc)
1111 return (0);
1112 break;
1113 }
1114 if ((name++)->w_wc == EOS)
1115 return (0);
1116 break;
1117 case M_SET:
1118 if (c.w_at != M_QUOTE) {
1119 k = *name++;
1120 if (k.w_at != c.w_at || k.w_wc != c.w_wc)
1121 return (0);
1122 break;
1123 }
1124 ok = 0;
1125 if ((k = *name++).w_wc == EOS)
1126 return (0);
1127 if ((negate_range = (pat->w_at == M_QUOTE &&
1128 pat->w_wc == M_NOT)) != 0)
1129 ++pat;
1130 while (((c = *pat++).w_at != M_QUOTE) ||
1131 c.w_wc != M_END) {
1132 if (c.w_at == M_QUOTE && c.w_wc == M_CLASS) {
1133 wcat_t cc;
1134
1135 cc.w_at = pat->w_at;
1136 cc.w_wc = pat->w_wc;
1137 if (iswctype(k.w_wc, cc.w_wc))
1138 ok = 1;
1139 ++pat;
1140 }
1141 if (pat->w_at == M_QUOTE &&
1142 pat->w_wc == M_RNG) {
1143 if (c.w_wc <= k.w_wc &&
1144 k.w_wc <= pat[1].w_wc)
1145 ok = 1;
1146 pat += 2;
1147 } else if (c.w_wc == k.w_wc)
1148 ok = 1;
1149 }
1150 if (ok == negate_range)
1151 return (0);
1152 break;
1153 default:
1154 k = *name++;
1155 if (k.w_at != c.w_at || k.w_wc != c.w_wc)
1156 return (0);
1157 break;
1158 }
1159 }
1160 return (name->w_wc == EOS);
1161 }
1162
1163 /*
1164 * Extended globfree() function, selected by #pragma redefine_extname
1165 * in glob.h with the external name _globfree_ext() .
1166 */
1167 void
1168 _globfree_ext(glob_t *pglob)
1169 {
1170 int i;
1171 char **pp;
1172
1173 if (pglob->gl_pathv != NULL) {
1174 pp = pglob->gl_pathv + pglob->gl_offs;
1175 for (i = pglob->gl_pathc; i--; ++pp)
1176 if (*pp)
1177 free(*pp);
1178 free(pglob->gl_pathv);
1179 pglob->gl_pathv = NULL;
1180 }
1181 if ((pglob->gl_flags & GLOB_KEEPSTAT) != 0 &&
1182 pglob->gl_statv != NULL) {
1183 for (i = 0; i < pglob->gl_pathc; i++) {
1184 if (pglob->gl_statv[i] != NULL)
1185 free(pglob->gl_statv[i]);
1186 }
1187 free(pglob->gl_statv);
1188 pglob->gl_statv = NULL;
1189 }
1190 }
1191
1192 static DIR *
1193 g_opendir(wcat_t *str, glob_t *pglob)
1194 {
1195 char buf[MAXPATHLEN];
1196
1197 if (str->w_wc == EOS)
1198 (void) strlcpy(buf, ".", sizeof (buf));
1199 else {
1200 if (g_Ctoc(str, buf, sizeof (buf)))
1201 return (NULL);
1202 }
1203
1204 if (pglob->gl_flags & GLOB_ALTDIRFUNC)
1205 return ((*pglob->gl_opendir)(buf));
1206
1207 return (opendir(buf));
1208 }
1209
1210 static int
1211 g_lstat(wcat_t *fn, struct stat *sb, glob_t *pglob)
1212 {
1213 char buf[MAXPATHLEN];
1214
1215 if (g_Ctoc(fn, buf, sizeof (buf)))
1216 return (-1);
1217 if (pglob->gl_flags & GLOB_ALTDIRFUNC)
1218 return ((*pglob->gl_lstat)(buf, sb));
1219 return (lstat(buf, sb));
1220 }
1221
1222 static int
1223 g_stat(wcat_t *fn, struct stat *sb, glob_t *pglob)
1224 {
1225 char buf[MAXPATHLEN];
1226
1227 if (g_Ctoc(fn, buf, sizeof (buf)))
1228 return (-1);
1229 if (pglob->gl_flags & GLOB_ALTDIRFUNC)
1230 return ((*pglob->gl_stat)(buf, sb));
1231 return (stat(buf, sb));
1232 }
1233
1234 static wcat_t *
1235 g_strchr(const wcat_t *str, wchar_t ch)
1236 {
1237 do {
1238 if (str->w_at == 0 && str->w_wc == ch)
1239 return ((wcat_t *)str);
1240 } while ((str++)->w_wc != EOS);
1241 return (NULL);
1242 }
1243
1244 static int
1245 g_Ctoc(const wcat_t *str, char *buf, uint_t len)
1246 {
1247 int n;
1248 wchar_t w;
1249
1250 while (len >= MB_LEN_MAX) {
1251 w = (str++)->w_wc;
1252 if ((n = wctomb(buf, w)) > 0) {
1253 len -= n;
1254 buf += n;
1255 }
1256 if (n < 0)
1257 break;
1258 if (w == EOS)
1259 return (0);
1260 }
1261 return (1);
1262 }
1263
1264 /* glob() function with legacy glob structure */
1265 int
1266 old_glob(const char *pattern, int flags, int (*errfunc)(const char *, int),
1267 old_glob_t *pglob)
1268 {
1269
1270 glob_t gl;
1271 int rv;
1272
1273 flags &= GLOB_POSIX;
1274
1275 (void) memset(&gl, 0, sizeof (gl));
1276
1277 /*
1278 * Copy all the members, old to new. There's
1279 * really no point in micro-optimizing the copying.
1280 * Other members are set to zero.
1281 */
1282 gl.gl_pathc = pglob->gl_pathc;
1283 gl.gl_pathv = pglob->gl_pathv;
1284 gl.gl_offs = pglob->gl_offs;
1285 gl.gl_pathp = pglob->gl_pathp;
1286 gl.gl_pathn = pglob->gl_pathn;
1287
1288 rv = _glob_ext(pattern, flags, errfunc, &gl);
1289
1290 /*
1291 * Copy all the members, new to old. There's
1292 * really no point in micro-optimizing the copying.
1293 */
1294 pglob->gl_pathc = gl.gl_pathc;
1295 pglob->gl_pathv = gl.gl_pathv;
1296 pglob->gl_offs = gl.gl_offs;
1297 pglob->gl_pathp = gl.gl_pathp;
1298 pglob->gl_pathn = gl.gl_pathn;
1299
1300 return (rv);
1301 }
1302
1303 /* globfree() function with legacy glob structure */
1304 void
1305 old_globfree(old_glob_t *pglob)
1306 {
1307 glob_t gl;
1308
1309 (void) memset(&gl, 0, sizeof (gl));
1310
1311 /*
1312 * Copy all the members, old to new. There's
1313 * really no point in micro-optimizing the copying.
1314 * Other members are set to zero.
1315 */
1316 gl.gl_pathc = pglob->gl_pathc;
1317 gl.gl_pathv = pglob->gl_pathv;
1318 gl.gl_offs = pglob->gl_offs;
1319 gl.gl_pathp = pglob->gl_pathp;
1320 gl.gl_pathn = pglob->gl_pathn;
1321
1322 _globfree_ext(&gl);
1323
1324 /*
1325 * Copy all the members, new to old. There's
1326 * really no point in micro-optimizing the copying.
1327 */
1328 pglob->gl_pathc = gl.gl_pathc;
1329 pglob->gl_pathv = gl.gl_pathv;
1330 pglob->gl_offs = gl.gl_offs;
1331 pglob->gl_pathp = gl.gl_pathp;
1332 pglob->gl_pathn = gl.gl_pathn;
1333
1334 }
1335
1336 /* End */