xref: /freebsd/contrib/tcsh/glob.c (revision d93a896e)
1 /*
2  * Copyright (c) 1989 The Regents of the University of California.
3  * All rights reserved.
4  *
5  * This code is derived from software contributed to Berkeley by
6  * Guido van Rossum.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the name of the University nor the names of its contributors
17  *    may be used to endorse or promote products derived from this software
18  *    without specific prior written permission.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  */
32 #if defined(LIBC_SCCS) && !defined(lint)
33 static char sccsid[] = "@(#)glob.c	5.12 (Berkeley) 6/24/91";
34 #endif /* LIBC_SCCS and not lint */
35 /*
36  * Glob: the interface is a superset of the one defined in POSIX 1003.2,
37  * draft 9.
38  *
39  * The [!...] convention to negate a range is supported (SysV, Posix, ksh).
40  *
41  * Optional extra services, controlled by flags not defined by POSIX:
42  *
43  * GLOB_QUOTE:
44  *	Escaping convention: \ inhibits any special meaning the following
45  *	character might have (except \ at end of string is retained).
46  * GLOB_MAGCHAR:
47  *	Set in gl_flags if pattern contained a globbing character.
48  * GLOB_ALTNOT:
49  *	Use ^ instead of ! for "not".
50  * gl_matchc:
51  *	Number of matches in the current invocation of glob.
52  */
53 
54 #ifdef WINNT_NATIVE
55 	#pragma warning(disable:4244)
56 #endif /* WINNT_NATIVE */
57 
58 #define Char __Char
59 #include "sh.h"
60 #include "glob.h"
61 
62 #ifndef HAVE_MBLEN
63 #undef mblen
64 #define mblen(_s,_n)	mbrlen((_s),(_n),NULL)
65 #endif
66 
67 #undef Char
68 #undef QUOTE
69 #undef TILDE
70 #undef META
71 #undef ismeta
72 #undef Strchr
73 
74 #ifndef S_ISDIR
75 #define S_ISDIR(a)	(((a) & S_IFMT) == S_IFDIR)
76 #endif
77 
78 #if !defined(S_ISLNK) && defined(S_IFLNK)
79 #define S_ISLNK(a)	(((a) & S_IFMT) == S_IFLNK)
80 #endif
81 
82 #if !defined(S_ISLNK) && !defined(lstat)
83 #define lstat stat
84 #endif
85 
86 typedef unsigned short Char;
87 
88 static	int	 glob1 		(Char *, glob_t *, int);
89 static	int	 glob2		(struct strbuf *, const Char *, glob_t *, int);
90 static	int	 glob3		(struct strbuf *, const Char *, const Char *,
91 				 const Char *, glob_t *, int);
92 static	void	 globextend	(const char *, glob_t *);
93 static	int	 match		(const char *, const Char *, const Char *,
94 				 int);
95 static	int	 compare	(const void *, const void *);
96 static 	DIR	*Opendir	(const char *);
97 #ifdef S_IFLNK
98 static	int	 Lstat		(const char *, struct stat *);
99 #endif
100 static	int	 Stat		(const char *, struct stat *sb);
101 static 	Char 	*Strchr		(Char *, int);
102 #ifdef DEBUG
103 static	void	 qprintf	(const Char *);
104 #endif
105 
106 #define	DOLLAR		'$'
107 #define	DOT		'.'
108 #define	EOS		'\0'
109 #define	LBRACKET	'['
110 #define	NOT		'!'
111 #define ALTNOT		'^'
112 #define	QUESTION	'?'
113 #define	QUOTE		'\\'
114 #define	RANGE		'-'
115 #define	RBRACKET	']'
116 #define	SEP		'/'
117 #define	STAR		'*'
118 #define	TILDE		'~'
119 #define	UNDERSCORE	'_'
120 
121 #define	M_META		0x8000
122 #define M_PROTECT	0x4000
123 #define	M_MASK		0xffff
124 #define	M_ASCII		0x00ff
125 
126 #define	LCHAR(c)	((c)&M_ASCII)
127 #define	META(c)		((c)|M_META)
128 #define	M_ALL		META('*')
129 #define	M_END		META(']')
130 #define	M_NOT		META('!')
131 #define	M_ALTNOT	META('^')
132 #define	M_ONE		META('?')
133 #define	M_RNG		META('-')
134 #define	M_SET		META('[')
135 #define	ismeta(c)	(((c)&M_META) != 0)
136 
137 int
138 globcharcoll(__Char c1, __Char c2, int cs)
139 {
140 #if defined(NLS) && defined(LC_COLLATE) && defined(HAVE_STRCOLL)
141 # if defined(WIDE_STRINGS)
142     wchar_t s1[2], s2[2];
143 
144     if (c1 == c2)
145 	return (0);
146     if (cs) {
147 	c1 = towlower(c1);
148 	c2 = towlower(c2);
149     } else {
150 #ifndef __FreeBSD__
151 	/* This should not be here, but I'll rather leave it in than engage in
152 	   a LC_COLLATE flamewar about a shell I don't use... */
153 	if (iswlower(c1) && iswupper(c2))
154 	    return (1);
155 	if (iswupper(c1) && iswlower(c2))
156 	    return (-1);
157 #endif
158     }
159     s1[0] = c1;
160     s2[0] = c2;
161     s1[1] = s2[1] = '\0';
162     return wcscoll(s1, s2);
163 # else /* not WIDE_STRINGS */
164     char s1[2], s2[2];
165 
166     if (c1 == c2)
167 	return (0);
168     /*
169      * From kevin lyda <kevin@suberic.net>:
170      * strcoll does not guarantee case sorting, so we pre-process now:
171      */
172     if (cs) {
173 	c1 = islower(c1) ? c1 : tolower(c1);
174 	c2 = islower(c2) ? c2 : tolower(c2);
175     } else {
176 	if (islower(c1) && isupper(c2))
177 	    return (1);
178 	if (isupper(c1) && islower(c2))
179 	    return (-1);
180     }
181     s1[0] = c1;
182     s2[0] = c2;
183     s1[1] = s2[1] = '\0';
184     return strcoll(s1, s2);
185 # endif
186 #else
187     return (c1 - c2);
188 #endif
189 }
190 
191 /*
192  * Need to dodge two kernel bugs:
193  * opendir("") != opendir(".")
194  * NAMEI_BUG: on plain files trailing slashes are ignored in some kernels.
195  *            POSIX specifies that they should be ignored in directories.
196  */
197 
198 static DIR *
199 Opendir(const char *str)
200 {
201 #if defined(hpux) || defined(__hpux)
202     struct stat st;
203 #endif
204 
205     if (!*str)
206 	return (opendir("."));
207 #if defined(hpux) || defined(__hpux)
208     /*
209      * Opendir on some device files hangs, so avoid it
210      */
211     if (stat(str, &st) == -1 || !S_ISDIR(st.st_mode))
212 	return NULL;
213 #endif
214     return opendir(str);
215 }
216 
217 #ifdef S_IFLNK
218 static int
219 Lstat(const char *fn, struct stat *sb)
220 {
221     int st;
222 
223     st = lstat(fn, sb);
224 # ifdef NAMEI_BUG
225     if (*fn != 0 && strend(fn)[-1] == '/' && !S_ISDIR(sb->st_mode))
226 	st = -1;
227 # endif	/* NAMEI_BUG */
228     return st;
229 }
230 #else
231 #define Lstat Stat
232 #endif /* S_IFLNK */
233 
234 static int
235 Stat(const char *fn, struct stat *sb)
236 {
237     int st;
238 
239     st = stat(fn, sb);
240 #ifdef NAMEI_BUG
241     if (*fn != 0 && strend(fn)[-1] == '/' && !S_ISDIR(sb->st_mode))
242 	st = -1;
243 #endif /* NAMEI_BUG */
244     return st;
245 }
246 
247 static Char *
248 Strchr(Char *str, int ch)
249 {
250     do
251 	if (*str == ch)
252 	    return (str);
253     while (*str++);
254     return (NULL);
255 }
256 
257 #ifdef DEBUG
258 static void
259 qprintf(const Char *s)
260 {
261     const Char *p;
262 
263     for (p = s; *p; p++)
264 	printf("%c", *p & 0xff);
265     printf("\n");
266     for (p = s; *p; p++)
267 	printf("%c", *p & M_PROTECT ? '"' : ' ');
268     printf("\n");
269     for (p = s; *p; p++)
270 	printf("%c", *p & M_META ? '_' : ' ');
271     printf("\n");
272 }
273 #endif /* DEBUG */
274 
275 static int
276 compare(const void *p, const void *q)
277 {
278 #if defined(NLS) && defined(HAVE_STRCOLL)
279     return (strcoll(*(char *const *) p, *(char *const *) q));
280 #else
281     return (strcmp(*(char *const *) p, *(char *const *) q));
282 #endif /* NLS && HAVE_STRCOLL */
283 }
284 
285 /*
286  * The main glob() routine: compiles the pattern (optionally processing
287  * quotes), calls glob1() to do the real pattern matching, and finally
288  * sorts the list (unless unsorted operation is requested).  Returns 0
289  * if things went well, nonzero if errors occurred.  It is not an error
290  * to find no matches.
291  */
292 int
293 glob(const char *pattern, int flags, int (*errfunc) (const char *, int),
294      glob_t *pglob)
295 {
296     int     err, oldpathc;
297     Char *bufnext, m_not;
298     const unsigned char *patnext;
299     int     c, not;
300     Char *qpatnext, *patbuf;
301     int     no_match;
302 
303     patnext = (const unsigned char *) pattern;
304     if (!(flags & GLOB_APPEND)) {
305 	pglob->gl_pathc = 0;
306 	pglob->gl_pathv = NULL;
307 	if (!(flags & GLOB_DOOFFS))
308 	    pglob->gl_offs = 0;
309     }
310     pglob->gl_flags = flags & ~GLOB_MAGCHAR;
311     pglob->gl_errfunc = errfunc;
312     oldpathc = pglob->gl_pathc;
313     pglob->gl_matchc = 0;
314 
315     if (pglob->gl_flags & GLOB_ALTNOT) {
316 	not = ALTNOT;
317 	m_not = M_ALTNOT;
318     }
319     else {
320 	not = NOT;
321 	m_not = M_NOT;
322     }
323 
324     patbuf = xmalloc((strlen(pattern) + 1) * sizeof(*patbuf));
325     bufnext = patbuf;
326 
327     no_match = *patnext == not;
328     if (no_match)
329 	patnext++;
330 
331     if (flags & GLOB_QUOTE) {
332 	/* Protect the quoted characters */
333 	while ((c = *patnext++) != EOS) {
334 #ifdef WIDE_STRINGS
335 	    int len;
336 
337 	    len = mblen((const char *)(patnext - 1), MB_LEN_MAX);
338 	    if (len == -1)
339 		TCSH_IGNORE(mblen(NULL, 0));
340 	    else if (len > 1) {
341 		*bufnext++ = (Char) c;
342 		while (--len != 0)
343 		    *bufnext++ = (Char) (*patnext++ | M_PROTECT);
344 	    } else
345 #endif /* WIDE_STRINGS */
346 	    if (c == QUOTE) {
347 		if ((c = *patnext++) == EOS) {
348 		    c = QUOTE;
349 		    --patnext;
350 		}
351 		*bufnext++ = (Char) (c | M_PROTECT);
352 	    }
353 	    else
354 		*bufnext++ = (Char) c;
355 	}
356     }
357     else
358 	while ((c = *patnext++) != EOS)
359 	    *bufnext++ = (Char) c;
360     *bufnext = EOS;
361 
362     bufnext = patbuf;
363     qpatnext = patbuf;
364     while ((c = *qpatnext++) != EOS) {
365 	switch (c) {
366 	case LBRACKET:
367 	    c = *qpatnext;
368 	    if (c == not)
369 		++qpatnext;
370 	    if (*qpatnext == EOS ||
371 		Strchr(qpatnext + 1, RBRACKET) == NULL) {
372 		*bufnext++ = LBRACKET;
373 		if (c == not)
374 		    --qpatnext;
375 		break;
376 	    }
377 	    pglob->gl_flags |= GLOB_MAGCHAR;
378 	    *bufnext++ = M_SET;
379 	    if (c == not)
380 		*bufnext++ = m_not;
381 	    c = *qpatnext++;
382 	    do {
383 		*bufnext++ = LCHAR(c);
384 		if (*qpatnext == RANGE &&
385 		    (c = qpatnext[1]) != RBRACKET) {
386 		    *bufnext++ = M_RNG;
387 		    *bufnext++ = LCHAR(c);
388 		    qpatnext += 2;
389 		}
390 	    } while ((c = *qpatnext++) != RBRACKET);
391 	    *bufnext++ = M_END;
392 	    break;
393 	case QUESTION:
394 	    pglob->gl_flags |= GLOB_MAGCHAR;
395 	    *bufnext++ = M_ONE;
396 	    break;
397 	case STAR:
398 	    pglob->gl_flags |= GLOB_MAGCHAR;
399 	    /* collapse adjacent stars to one [or three if globstar],
400 	     * to avoid exponential behavior
401 	     */
402 	    if (bufnext == patbuf || bufnext[-1] != M_ALL ||
403 	       ((flags & GLOB_STAR) != 0 &&
404 		 (bufnext - 1 == patbuf || bufnext[-2] != M_ALL ||
405 		 bufnext - 2 == patbuf || bufnext[-3] != M_ALL)))
406 		*bufnext++ = M_ALL;
407 	    break;
408 	default:
409 	    *bufnext++ = LCHAR(c);
410 	    break;
411 	}
412     }
413     *bufnext = EOS;
414 #ifdef DEBUG
415     qprintf(patbuf);
416 #endif
417 
418     if ((err = glob1(patbuf, pglob, no_match)) != 0) {
419 	xfree(patbuf);
420 	return (err);
421     }
422 
423     /*
424      * If there was no match we are going to append the pattern
425      * if GLOB_NOCHECK was specified or if GLOB_NOMAGIC was specified
426      * and the pattern did not contain any magic characters
427      * GLOB_NOMAGIC is there just for compatibility with csh.
428      */
429     if (pglob->gl_pathc == oldpathc &&
430 	((flags & GLOB_NOCHECK) ||
431 	 ((flags & GLOB_NOMAGIC) && !(pglob->gl_flags & GLOB_MAGCHAR)))) {
432 	if (!(flags & GLOB_QUOTE))
433 	    globextend(pattern, pglob);
434 	else {
435 	    char *copy, *dest;
436 	    const char *src;
437 
438 	    /* copy pattern, interpreting quotes */
439 	    copy = xmalloc(strlen(pattern) + 1);
440 	    dest = copy;
441 	    src = pattern;
442 	    while (*src != EOS) {
443 		/* Don't interpret quotes. The spec does not say we should do */
444 		if (*src == QUOTE) {
445 		    if (*++src == EOS)
446 			--src;
447 		}
448 		*dest++ = *src++;
449 	    }
450 	    *dest = EOS;
451 	    globextend(copy, pglob);
452 	    xfree(copy);
453 	}
454 	xfree(patbuf);
455 	return 0;
456     }
457     else if (!(flags & GLOB_NOSORT) && (pglob->gl_pathc != oldpathc))
458 	qsort(pglob->gl_pathv + pglob->gl_offs + oldpathc,
459 	      pglob->gl_pathc - oldpathc, sizeof(char *), compare);
460     xfree(patbuf);
461     return (0);
462 }
463 
464 static int
465 glob1(Char *pattern, glob_t *pglob, int no_match)
466 {
467     struct strbuf pathbuf = strbuf_INIT;
468     int err;
469 
470     /*
471      * a null pathname is invalid -- POSIX 1003.1 sect. 2.4.
472      */
473     if (*pattern == EOS)
474 	return (0);
475     err = glob2(&pathbuf, pattern, pglob, no_match);
476     xfree(pathbuf.s);
477     return err;
478 }
479 
480 /*
481  * functions glob2 and glob3 are mutually recursive; there is one level
482  * of recursion for each segment in the pattern that contains one or
483  * more meta characters.
484  */
485 static int
486 glob2(struct strbuf *pathbuf, const Char *pattern, glob_t *pglob, int no_match)
487 {
488     struct stat sbuf;
489     int anymeta;
490     const Char *p;
491     size_t orig_len;
492 
493     /*
494      * loop over pattern segments until end of pattern or until segment with
495      * meta character found.
496      */
497     anymeta = 0;
498     for (;;) {
499 	if (*pattern == EOS) {	/* end of pattern? */
500 	    strbuf_terminate(pathbuf);
501 
502 	    if (Lstat(pathbuf->s, &sbuf))
503 		return (0);
504 
505 	    if (((pglob->gl_flags & GLOB_MARK) &&
506 		 pathbuf->s[pathbuf->len - 1] != SEP) &&
507 		(S_ISDIR(sbuf.st_mode)
508 #ifdef S_IFLNK
509 		 || (S_ISLNK(sbuf.st_mode) &&
510 		     (Stat(pathbuf->s, &sbuf) == 0) &&
511 		     S_ISDIR(sbuf.st_mode))
512 #endif
513 		 )) {
514 		strbuf_append1(pathbuf, SEP);
515 		strbuf_terminate(pathbuf);
516 	    }
517 	    ++pglob->gl_matchc;
518 	    globextend(pathbuf->s, pglob);
519 	    return 0;
520 	}
521 
522 	/* find end of next segment, tentatively copy to pathbuf */
523 	p = pattern;
524 	orig_len = pathbuf->len;
525 	while (*p != EOS && *p != SEP) {
526 	    if (ismeta(*p))
527 		anymeta = 1;
528 	    strbuf_append1(pathbuf, *p++);
529 	}
530 
531 	if (!anymeta) {		/* no expansion, do next segment */
532 	    pattern = p;
533 	    while (*pattern == SEP)
534 		strbuf_append1(pathbuf, *pattern++);
535 	}
536 	else {			/* need expansion, recurse */
537 	    pathbuf->len = orig_len;
538 	    return (glob3(pathbuf, pattern, p, pattern, pglob, no_match));
539 	}
540     }
541     /* NOTREACHED */
542 }
543 
544 static size_t
545 One_Char_mbtowc(__Char *pwc, const Char *s, size_t n)
546 {
547 #ifdef WIDE_STRINGS
548     char buf[MB_LEN_MAX], *p;
549 
550     if (n > MB_LEN_MAX)
551 	n = MB_LEN_MAX;
552     p = buf;
553     while (p < buf + n && (*p++ = LCHAR(*s++)) != 0)
554 	;
555     return one_mbtowc(pwc, buf, n);
556 #else
557     *pwc = *s & CHAR;
558     return 1;
559 #endif
560 }
561 
562 static int
563 glob3(struct strbuf *pathbuf, const Char *pattern, const Char *restpattern,
564       const Char *pglobstar, glob_t *pglob, int no_match)
565 {
566     DIR    *dirp;
567     struct dirent *dp;
568     struct stat sbuf;
569     int     err;
570     Char m_not = (pglob->gl_flags & GLOB_ALTNOT) ? M_ALTNOT : M_NOT;
571     size_t orig_len;
572     int globstar = 0;
573     int chase_symlinks = 0;
574     const Char *termstar = NULL;
575 
576     strbuf_terminate(pathbuf);
577     orig_len = pathbuf->len;
578     errno = err = 0;
579 
580     while (pglobstar < restpattern) {
581 	__Char wc;
582 	size_t width = One_Char_mbtowc(&wc, pglobstar, MB_LEN_MAX);
583 	if ((pglobstar[0] & M_MASK) == M_ALL &&
584 	    (pglobstar[width] & M_MASK) == M_ALL) {
585 	    globstar = 1;
586 	    chase_symlinks = (pglobstar[2 * width] & M_MASK) == M_ALL;
587 	    termstar = pglobstar + (2 + chase_symlinks) * width;
588 	    break;
589 	}
590         pglobstar += width;
591     }
592 
593     if (globstar) {
594 	err = pglobstar==pattern && termstar==restpattern ?
595 		*restpattern == EOS ?
596 		glob2(pathbuf, restpattern - 1, pglob, no_match) :
597 		glob2(pathbuf, restpattern + 1, pglob, no_match) :
598 		glob3(pathbuf, pattern, restpattern, termstar, pglob, no_match);
599 	if (err)
600 	    return err;
601 	pathbuf->len = orig_len;
602 	strbuf_terminate(pathbuf);
603     }
604 
605     if (*pathbuf->s && (Lstat(pathbuf->s, &sbuf) || !S_ISDIR(sbuf.st_mode)
606 #ifdef S_IFLINK
607 	     && ((globstar && !chase_symlinks) || !S_ISLNK(sbuf.st_mode))
608 #endif
609 	))
610 	return 0;
611 
612     if (!(dirp = Opendir(pathbuf->s))) {
613 	/* todo: don't call for ENOENT or ENOTDIR? */
614 	if ((pglob->gl_errfunc && (*pglob->gl_errfunc) (pathbuf->s, errno)) ||
615 	    (pglob->gl_flags & GLOB_ERR))
616 	    return (GLOB_ABEND);
617 	else
618 	    return (0);
619     }
620 
621     /* search directory for matching names */
622     while ((dp = readdir(dirp)) != NULL) {
623 	/* initial DOT must be matched literally */
624 	if (dp->d_name[0] == DOT && *pattern != DOT)
625 	    if (!(pglob->gl_flags & GLOB_DOT) || !dp->d_name[1] ||
626 		(dp->d_name[1] == DOT && !dp->d_name[2]))
627 		continue; /*unless globdot and not . or .. */
628 	pathbuf->len = orig_len;
629 	strbuf_append(pathbuf, dp->d_name);
630 	strbuf_terminate(pathbuf);
631 
632 	if (globstar) {
633 #ifdef S_IFLNK
634 	    if (!chase_symlinks &&
635 		(Lstat(pathbuf->s, &sbuf) || S_ISLNK(sbuf.st_mode)))
636 		    continue;
637 #endif
638 	    if (match(pathbuf->s + orig_len, pattern, termstar,
639 		(int)m_not) == no_match)
640 		    continue;
641 	    strbuf_append1(pathbuf, SEP);
642 	    strbuf_terminate(pathbuf);
643 	    if ((err = glob2(pathbuf, pglobstar, pglob, no_match)) != 0)
644 		break;
645 	} else {
646 	    if (match(pathbuf->s + orig_len, pattern, restpattern,
647 		(int) m_not) == no_match)
648 		continue;
649 	    if ((err = glob2(pathbuf, restpattern, pglob, no_match)) != 0)
650 		break;
651 	}
652     }
653     /* todo: check error from readdir? */
654     closedir(dirp);
655     return (err);
656 }
657 
658 
659 /*
660  * Extend the gl_pathv member of a glob_t structure to accomodate a new item,
661  * add the new item, and update gl_pathc.
662  *
663  * This assumes the BSD realloc, which only copies the block when its size
664  * crosses a power-of-two boundary; for v7 realloc, this would cause quadratic
665  * behavior.
666  *
667  * Return 0 if new item added, error code if memory couldn't be allocated.
668  *
669  * Invariant of the glob_t structure:
670  *	Either gl_pathc is zero and gl_pathv is NULL; or gl_pathc > 0 and
671  *	 gl_pathv points to (gl_offs + gl_pathc + 1) items.
672  */
673 static void
674 globextend(const char *path, glob_t *pglob)
675 {
676     char **pathv;
677     int i;
678     size_t newsize;
679 
680     newsize = sizeof(*pathv) * (2 + pglob->gl_pathc + pglob->gl_offs);
681     pathv = xrealloc(pglob->gl_pathv, newsize);
682 
683     if (pglob->gl_pathv == NULL && pglob->gl_offs > 0) {
684 	/* first time around -- clear initial gl_offs items */
685 	pathv += pglob->gl_offs;
686 	for (i = pglob->gl_offs; --i >= 0;)
687 	    *--pathv = NULL;
688     }
689     pglob->gl_pathv = pathv;
690 
691     pathv[pglob->gl_offs + pglob->gl_pathc++] = strsave(path);
692     pathv[pglob->gl_offs + pglob->gl_pathc] = NULL;
693 }
694 
695 /*
696  * pattern matching function for filenames.  Each occurrence of the *
697  * pattern causes a recursion level.
698  */
699 static  int
700 match(const char *name, const Char *pat, const Char *patend, int m_not)
701 {
702     int ok, negate_range;
703     Char c;
704 
705     while (pat < patend) {
706 	size_t lwk;
707 	__Char wc, wk;
708 
709 	c = *pat; /* Only for M_MASK bits */
710 	pat += One_Char_mbtowc(&wc, pat, MB_LEN_MAX);
711 	lwk = one_mbtowc(&wk, name, MB_LEN_MAX);
712 	switch (c & M_MASK) {
713 	case M_ALL:
714 	    while (pat < patend && (*pat & M_MASK) == M_ALL)  /* eat consecutive '*' */
715 		pat += One_Char_mbtowc(&wc, pat, MB_LEN_MAX);
716 	    if (pat == patend)
717 	        return (1);
718 	    while (!match(name, pat, patend, m_not)) {
719 		if (*name == EOS)
720 		    return (0);
721 		name += lwk;
722 		lwk = one_mbtowc(&wk, name, MB_LEN_MAX);
723 	    }
724 	    return (1);
725 	case M_ONE:
726 	    if (*name == EOS)
727 		return (0);
728 	    name += lwk;
729 	    break;
730 	case M_SET:
731 	    ok = 0;
732 	    if (*name == EOS)
733 		return (0);
734 	    name += lwk;
735 	    if ((negate_range = ((*pat & M_MASK) == m_not)) != 0)
736 		++pat;
737 	    while ((*pat & M_MASK) != M_END) {
738 		pat += One_Char_mbtowc(&wc, pat, MB_LEN_MAX);
739 		if ((*pat & M_MASK) == M_RNG) {
740 		    __Char wc2;
741 
742 		    pat++;
743 		    pat += One_Char_mbtowc(&wc2, pat, MB_LEN_MAX);
744 		    if (globcharcoll(wc, wk, 0) <= 0 &&
745 			globcharcoll(wk, wc2, 0) <= 0)
746 			ok = 1;
747 		} else if (wc == wk)
748 		    ok = 1;
749 	    }
750 	    pat += One_Char_mbtowc(&wc, pat, MB_LEN_MAX);
751 	    if (ok == negate_range)
752 		return (0);
753 	    break;
754 	default:
755 	    if (*name == EOS || samecase(wk) != samecase(wc))
756 		return (0);
757 	    name += lwk;
758 	    break;
759 	}
760     }
761     return (*name == EOS);
762 }
763 
764 /* free allocated data belonging to a glob_t structure */
765 void
766 globfree(glob_t *pglob)
767 {
768     int i;
769     char **pp;
770 
771     if (pglob->gl_pathv != NULL) {
772 	pp = pglob->gl_pathv + pglob->gl_offs;
773 	for (i = pglob->gl_pathc; i--; ++pp)
774 	    if (*pp)
775 		xfree(*pp), *pp = NULL;
776 	xfree(pglob->gl_pathv), pglob->gl_pathv = NULL;
777     }
778 }
779