]>
Commit | Line | Data |
---|---|---|
224c7076 A |
1 | /* |
2 | * Copyright (c) 1989, 1993 | |
3 | * The Regents of the University of California. 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. | |
224c7076 A |
16 | * 4. 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 | ||
33 | #if defined(LIBC_SCCS) && !defined(lint) | |
34 | static char sccsid[] = "@(#)glob.c 8.3 (Berkeley) 10/13/93"; | |
35 | #endif /* LIBC_SCCS and not lint */ | |
36 | #include <sys/cdefs.h> | |
fbd86d4c | 37 | __FBSDID("$FreeBSD: src/lib/libc/gen/glob.c,v 1.28 2010/05/12 17:44:00 gordon Exp $"); |
224c7076 A |
38 | |
39 | #include "xlocale_private.h" | |
40 | ||
41 | /* | |
42 | * glob(3) -- a superset of the one defined in POSIX 1003.2. | |
43 | * | |
44 | * The [!...] convention to negate a range is supported (SysV, Posix, ksh). | |
45 | * | |
46 | * Optional extra services, controlled by flags not defined by POSIX: | |
47 | * | |
48 | * GLOB_QUOTE: | |
49 | * Escaping convention: \ inhibits any special meaning the following | |
50 | * character might have (except \ at end of string is retained). | |
51 | * GLOB_MAGCHAR: | |
52 | * Set in gl_flags if pattern contained a globbing character. | |
53 | * GLOB_NOMAGIC: | |
54 | * Same as GLOB_NOCHECK, but it will only append pattern if it did | |
55 | * not contain any magic characters. [Used in csh style globbing] | |
56 | * GLOB_ALTDIRFUNC: | |
57 | * Use alternately specified directory access functions. | |
58 | * GLOB_TILDE: | |
59 | * expand ~user/foo to the /home/dir/of/user/foo | |
60 | * GLOB_BRACE: | |
61 | * expand {1,2}{a,b} to 1a 1b 2a 2b | |
62 | * gl_matchc: | |
63 | * Number of matches in the current invocation of glob. | |
64 | */ | |
65 | ||
66 | /* | |
67 | * Some notes on multibyte character support: | |
68 | * 1. Patterns with illegal byte sequences match nothing - even if | |
69 | * GLOB_NOCHECK is specified. | |
70 | * 2. Illegal byte sequences in filenames are handled by treating them as | |
71 | * single-byte characters with a value of the first byte of the sequence | |
72 | * cast to wchar_t. | |
73 | * 3. State-dependent encodings are not currently supported. | |
74 | */ | |
75 | ||
76 | #include <sys/param.h> | |
77 | #include <sys/stat.h> | |
78 | ||
79 | #include <ctype.h> | |
80 | #include <dirent.h> | |
81 | #include <errno.h> | |
82 | #include <glob.h> | |
83 | #include <limits.h> | |
84 | #include <pwd.h> | |
85 | #include <stdint.h> | |
86 | #include <stdio.h> | |
87 | #include <stdlib.h> | |
88 | #include <string.h> | |
89 | #include <unistd.h> | |
90 | #include <wchar.h> | |
91 | ||
92 | #include "collate.h" | |
93 | ||
fbd86d4c A |
94 | #define GLOB_LIMIT_STRING 65536 /* number of readdirs */ |
95 | #define GLOB_LIMIT_STAT 128 /* number of stat system calls */ | |
96 | #define GLOB_LIMIT_READDIR 16384 /* total buffer size of path strings */ | |
97 | #define GLOB_LIMIT_PATH 1024 /* number of path elements */ | |
98 | #define GLOB_LIMIT_BRACE 128 /* Number of brace calls */ | |
99 | ||
100 | struct glob_limit { | |
101 | size_t l_string; | |
102 | size_t l_stat; | |
103 | size_t l_readdir; | |
104 | size_t l_brace; | |
105 | }; | |
106 | ||
224c7076 A |
107 | #define DOLLAR '$' |
108 | #define DOT '.' | |
109 | #define EOS '\0' | |
110 | #define LBRACKET '[' | |
111 | #define NOT '!' | |
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 | #define LBRACE '{' | |
121 | #define RBRACE '}' | |
122 | #define SLASH '/' | |
123 | #define COMMA ',' | |
124 | ||
125 | #ifndef DEBUG | |
126 | ||
127 | #define M_QUOTE 0x8000000000ULL | |
128 | #define M_PROTECT 0x4000000000ULL | |
129 | #define M_MASK 0xffffffffffULL | |
130 | #define M_CHAR 0x00ffffffffULL | |
131 | ||
132 | typedef uint_fast64_t Char; | |
133 | ||
134 | #else | |
135 | ||
136 | #define M_QUOTE 0x80 | |
137 | #define M_PROTECT 0x40 | |
138 | #define M_MASK 0xff | |
139 | #define M_CHAR 0x7f | |
140 | ||
141 | typedef char Char; | |
142 | ||
143 | #endif | |
144 | ||
145 | ||
146 | #define CHAR(c) ((Char)((c)&M_CHAR)) | |
147 | #define META(c) ((Char)((c)|M_QUOTE)) | |
148 | #define M_ALL META('*') | |
149 | #define M_END META(']') | |
150 | #define M_NOT META('!') | |
151 | #define M_ONE META('?') | |
152 | #define M_RNG META('-') | |
153 | #define M_SET META('[') | |
154 | #define ismeta(c) (((c)&M_QUOTE) != 0) | |
155 | ||
224c7076 | 156 | |
b5d655f7 | 157 | #define compare __gl_compare |
224c7076 | 158 | #define g_Ctoc __gl_g_Ctoc |
b5d655f7 | 159 | #define g_strchr __gl_g_strchr |
224c7076 | 160 | #define globextend __gl_globextend |
b5d655f7 A |
161 | #define globtilde __gl_globtilde |
162 | #define match __gl_match | |
fbd86d4c A |
163 | __private_extern__ int compare(const void *, const void *); |
164 | __private_extern__ int g_Ctoc(const Char *, char *, size_t, locale_t); | |
165 | __private_extern__ const Char *g_strchr(const Char *, wchar_t); | |
166 | __private_extern__ int globextend(const Char *, glob_t *, struct glob_limit *, locale_t); | |
b5d655f7 A |
167 | __private_extern__ const Char * |
168 | globtilde(const Char *, Char *, size_t, glob_t *); | |
fbd86d4c | 169 | __private_extern__ int match(Char *, Char *, Char *, locale_t); |
224c7076 | 170 | |
224c7076 | 171 | |
b5d655f7 | 172 | static int g_lstat(Char *, struct stat *, glob_t *, locale_t); |
224c7076 | 173 | static DIR *g_opendir(Char *, glob_t *, locale_t); |
224c7076 A |
174 | #ifdef notdef |
175 | static Char *g_strcat(Char *, const Char *); | |
176 | #endif | |
b5d655f7 | 177 | static int g_stat(Char *, struct stat *, glob_t *, locale_t); |
fbd86d4c A |
178 | static int glob0(const Char *, glob_t *, struct glob_limit *, locale_t); |
179 | static int glob1(Char *, glob_t *, struct glob_limit *, locale_t); | |
180 | static int glob2(Char *, Char *, Char *, Char *, glob_t *, struct glob_limit *, locale_t); | |
181 | static int glob3(Char *, Char *, Char *, Char *, Char *, glob_t *, struct glob_limit *, locale_t); | |
182 | static int globexp1(const Char *, glob_t *, struct glob_limit *, locale_t); | |
183 | static int globexp2(const Char *, const Char *, glob_t *, int *, struct glob_limit *, locale_t); | |
224c7076 A |
184 | #ifdef DEBUG |
185 | static void qprintf(const char *, Char *); | |
186 | #endif | |
224c7076 | 187 | |
34e8f829 | 188 | static int |
fbd86d4c | 189 | __glob(const char *pattern, glob_t *pglob) |
224c7076 | 190 | { |
fbd86d4c A |
191 | const char *patnext; |
192 | struct glob_limit limit = { 0, 0, 0, 0 }; | |
224c7076 A |
193 | Char *bufnext, *bufend, patbuf[MAXPATHLEN], prot; |
194 | mbstate_t mbs; | |
195 | wchar_t wc; | |
196 | size_t clen; | |
197 | locale_t loc = __current_locale(); | |
198 | int mb_cur_max = MB_CUR_MAX_L(loc); | |
199 | ||
fbd86d4c | 200 | patnext = pattern; |
34e8f829 | 201 | if (!(pglob->gl_flags & GLOB_APPEND)) { |
224c7076 A |
202 | pglob->gl_pathc = 0; |
203 | pglob->gl_pathv = NULL; | |
34e8f829 | 204 | if (!(pglob->gl_flags & GLOB_DOOFFS)) |
224c7076 A |
205 | pglob->gl_offs = 0; |
206 | } | |
224c7076 A |
207 | pglob->gl_matchc = 0; |
208 | ||
209 | bufnext = patbuf; | |
210 | bufend = bufnext + MAXPATHLEN - 1; | |
34e8f829 | 211 | if (pglob->gl_flags & GLOB_NOESCAPE) { |
224c7076 A |
212 | memset(&mbs, 0, sizeof(mbs)); |
213 | while (bufend - bufnext >= mb_cur_max) { | |
fbd86d4c | 214 | clen = mbrtowc_l(&wc, patnext, MB_LEN_MAX, &mbs, loc); |
224c7076 A |
215 | if (clen == (size_t)-1 || clen == (size_t)-2) |
216 | return (GLOB_NOMATCH); | |
217 | else if (clen == 0) | |
218 | break; | |
219 | *bufnext++ = wc; | |
220 | patnext += clen; | |
221 | } | |
222 | } else { | |
223 | /* Protect the quoted characters. */ | |
224 | memset(&mbs, 0, sizeof(mbs)); | |
225 | while (bufend - bufnext >= mb_cur_max) { | |
226 | if (*patnext == QUOTE) { | |
227 | if (*++patnext == EOS) { | |
228 | *bufnext++ = QUOTE | M_PROTECT; | |
229 | continue; | |
230 | } | |
231 | prot = M_PROTECT; | |
232 | } else | |
233 | prot = 0; | |
fbd86d4c | 234 | clen = mbrtowc_l(&wc, patnext, MB_LEN_MAX, &mbs, loc); |
224c7076 A |
235 | if (clen == (size_t)-1 || clen == (size_t)-2) |
236 | return (GLOB_NOMATCH); | |
237 | else if (clen == 0) | |
238 | break; | |
239 | *bufnext++ = wc | prot; | |
240 | patnext += clen; | |
241 | } | |
242 | } | |
243 | *bufnext = EOS; | |
244 | ||
34e8f829 | 245 | if (pglob->gl_flags & GLOB_BRACE) |
224c7076 A |
246 | return globexp1(patbuf, pglob, &limit, loc); |
247 | else | |
248 | return glob0(patbuf, pglob, &limit, loc); | |
249 | } | |
250 | ||
34e8f829 | 251 | int |
fbd86d4c | 252 | glob(const char *pattern, int flags, int (*errfunc)(const char *, int), glob_t *pglob) |
34e8f829 A |
253 | { |
254 | #ifdef __BLOCKS__ | |
255 | pglob->gl_flags = flags & ~(GLOB_MAGCHAR | _GLOB_ERR_BLOCK); | |
256 | #else /* !__BLOCKS__ */ | |
257 | pglob->gl_flags = flags & ~GLOB_MAGCHAR; | |
258 | #endif /* __BLOCKS__ */ | |
259 | pglob->gl_errfunc = errfunc; | |
260 | return __glob(pattern, pglob); | |
261 | } | |
262 | ||
263 | #ifdef __BLOCKS__ | |
264 | int | |
fbd86d4c | 265 | glob_b(const char *pattern, int flags, int (^errblk)(const char *, int), glob_t *pglob) |
34e8f829 A |
266 | { |
267 | pglob->gl_flags = flags & ~GLOB_MAGCHAR; | |
268 | pglob->gl_flags |= _GLOB_ERR_BLOCK; | |
269 | pglob->gl_errblk = errblk; | |
270 | return __glob(pattern, pglob); | |
271 | } | |
272 | #endif /* __BLOCKS__ */ | |
273 | ||
224c7076 A |
274 | /* |
275 | * Expand recursively a glob {} pattern. When there is no more expansion | |
276 | * invoke the standard globbing routine to glob the rest of the magic | |
277 | * characters | |
278 | */ | |
b5d655f7 | 279 | static int |
fbd86d4c | 280 | globexp1(const Char *pattern, glob_t *pglob, struct glob_limit *limit, locale_t loc) |
224c7076 A |
281 | { |
282 | const Char* ptr = pattern; | |
283 | int rv; | |
284 | ||
fbd86d4c A |
285 | if ((pglob->gl_flags & GLOB_LIMIT) && |
286 | limit->l_brace++ >= GLOB_LIMIT_BRACE) { | |
287 | errno = 0; | |
288 | return GLOB_NOSPACE; | |
289 | } | |
290 | ||
224c7076 A |
291 | /* Protect a single {}, for find(1), like csh */ |
292 | if (pattern[0] == LBRACE && pattern[1] == RBRACE && pattern[2] == EOS) | |
293 | return glob0(pattern, pglob, limit, loc); | |
294 | ||
fbd86d4c | 295 | while ((ptr = g_strchr(ptr, LBRACE)) != NULL) |
224c7076 A |
296 | if (!globexp2(ptr, pattern, pglob, &rv, limit, loc)) |
297 | return rv; | |
298 | ||
299 | return glob0(pattern, pglob, limit, loc); | |
300 | } | |
301 | ||
302 | ||
303 | /* | |
304 | * Recursive brace globbing helper. Tries to expand a single brace. | |
305 | * If it succeeds then it invokes globexp1 with the new pattern. | |
306 | * If it fails then it tries to glob the rest of the pattern and returns. | |
307 | */ | |
308 | static int | |
fbd86d4c | 309 | globexp2(const Char *ptr, const Char *pattern, glob_t *pglob, int *rv, struct glob_limit *limit, locale_t loc) |
224c7076 A |
310 | { |
311 | int i; | |
312 | Char *lm, *ls; | |
fbd86d4c | 313 | const Char *pe, *pm, *pm1, *pl; |
224c7076 A |
314 | Char patbuf[MAXPATHLEN]; |
315 | ||
316 | /* copy part up to the brace */ | |
317 | for (lm = patbuf, pm = pattern; pm != ptr; *lm++ = *pm++) | |
318 | continue; | |
319 | *lm = EOS; | |
320 | ls = lm; | |
321 | ||
322 | /* Find the balanced brace */ | |
323 | for (i = 0, pe = ++ptr; *pe; pe++) | |
324 | if (*pe == LBRACKET) { | |
325 | /* Ignore everything between [] */ | |
326 | for (pm = pe++; *pe != RBRACKET && *pe != EOS; pe++) | |
327 | continue; | |
328 | if (*pe == EOS) { | |
329 | /* | |
330 | * We could not find a matching RBRACKET. | |
331 | * Ignore and just look for RBRACE | |
332 | */ | |
333 | pe = pm; | |
334 | } | |
335 | } | |
336 | else if (*pe == LBRACE) | |
337 | i++; | |
338 | else if (*pe == RBRACE) { | |
339 | if (i == 0) | |
340 | break; | |
341 | i--; | |
342 | } | |
343 | ||
344 | /* Non matching braces; just glob the pattern */ | |
345 | if (i != 0 || *pe == EOS) { | |
346 | *rv = glob0(patbuf, pglob, limit, loc); | |
347 | return 0; | |
348 | } | |
349 | ||
350 | for (i = 0, pl = pm = ptr; pm <= pe; pm++) | |
351 | switch (*pm) { | |
352 | case LBRACKET: | |
353 | /* Ignore everything between [] */ | |
fbd86d4c | 354 | for (pm1 = pm++; *pm != RBRACKET && *pm != EOS; pm++) |
224c7076 A |
355 | continue; |
356 | if (*pm == EOS) { | |
357 | /* | |
358 | * We could not find a matching RBRACKET. | |
359 | * Ignore and just look for RBRACE | |
360 | */ | |
fbd86d4c | 361 | pm = pm1; |
224c7076 A |
362 | } |
363 | break; | |
364 | ||
365 | case LBRACE: | |
366 | i++; | |
367 | break; | |
368 | ||
369 | case RBRACE: | |
370 | if (i) { | |
371 | i--; | |
372 | break; | |
373 | } | |
374 | /* FALLTHROUGH */ | |
375 | case COMMA: | |
376 | if (i && *pm == COMMA) | |
377 | break; | |
378 | else { | |
379 | /* Append the current string */ | |
380 | for (lm = ls; (pl < pm); *lm++ = *pl++) | |
381 | continue; | |
382 | /* | |
383 | * Append the rest of the pattern after the | |
384 | * closing brace | |
385 | */ | |
386 | for (pl = pe + 1; (*lm++ = *pl++) != EOS;) | |
387 | continue; | |
388 | ||
389 | /* Expand the current pattern */ | |
390 | #ifdef DEBUG | |
391 | qprintf("globexp2:", patbuf); | |
392 | #endif | |
393 | *rv = globexp1(patbuf, pglob, limit, loc); | |
394 | ||
395 | /* move after the comma, to the next string */ | |
396 | pl = pm + 1; | |
397 | } | |
398 | break; | |
399 | ||
400 | default: | |
401 | break; | |
402 | } | |
403 | *rv = 0; | |
404 | return 0; | |
405 | } | |
406 | ||
407 | ||
408 | ||
b5d655f7 | 409 | #ifndef BUILDING_VARIANT |
224c7076 A |
410 | /* |
411 | * expand tilde from the passwd file. | |
412 | */ | |
b5d655f7 | 413 | __private_extern__ const Char * |
fbd86d4c | 414 | globtilde(const Char *pattern, Char *patbuf, size_t patbuf_len, glob_t *pglob) |
224c7076 A |
415 | { |
416 | struct passwd *pwd; | |
417 | char *h; | |
418 | const Char *p; | |
419 | Char *b, *eb; | |
420 | ||
421 | if (*pattern != TILDE || !(pglob->gl_flags & GLOB_TILDE)) | |
422 | return pattern; | |
423 | ||
424 | /* | |
425 | * Copy up to the end of the string or / | |
426 | */ | |
427 | eb = &patbuf[patbuf_len - 1]; | |
428 | for (p = pattern + 1, h = (char *) patbuf; | |
429 | h < (char *)eb && *p && *p != SLASH; *h++ = *p++) | |
430 | continue; | |
431 | ||
432 | *h = EOS; | |
433 | ||
434 | if (((char *) patbuf)[0] == EOS) { | |
435 | /* | |
436 | * handle a plain ~ or ~/ by expanding $HOME first (iff | |
437 | * we're not running setuid or setgid) and then trying | |
438 | * the password file | |
439 | */ | |
440 | if (issetugid() != 0 || | |
441 | (h = getenv("HOME")) == NULL) { | |
442 | if (((h = getlogin()) != NULL && | |
443 | (pwd = getpwnam(h)) != NULL) || | |
444 | (pwd = getpwuid(getuid())) != NULL) | |
445 | h = pwd->pw_dir; | |
446 | else | |
447 | return pattern; | |
448 | } | |
449 | } | |
450 | else { | |
451 | /* | |
452 | * Expand a ~user | |
453 | */ | |
454 | if ((pwd = getpwnam((char*) patbuf)) == NULL) | |
455 | return pattern; | |
456 | else | |
457 | h = pwd->pw_dir; | |
458 | } | |
459 | ||
460 | /* Copy the home directory */ | |
461 | for (b = patbuf; b < eb && *h; *b++ = *h++) | |
462 | continue; | |
463 | ||
464 | /* Append the rest of the pattern */ | |
465 | while (b < eb && (*b++ = *p++) != EOS) | |
466 | continue; | |
467 | *b = EOS; | |
468 | ||
469 | return patbuf; | |
470 | } | |
b5d655f7 | 471 | #endif /* BUILDING_VARIANT */ |
224c7076 A |
472 | |
473 | ||
474 | /* | |
475 | * The main glob() routine: compiles the pattern (optionally processing | |
476 | * quotes), calls glob1() to do the real pattern matching, and finally | |
477 | * sorts the list (unless unsorted operation is requested). Returns 0 | |
478 | * if things went well, nonzero if errors occurred. | |
479 | */ | |
b5d655f7 | 480 | static int |
fbd86d4c | 481 | glob0(const Char *pattern, glob_t *pglob, struct glob_limit *limit, locale_t loc) |
224c7076 A |
482 | { |
483 | const Char *qpatnext; | |
fbd86d4c A |
484 | int err; |
485 | size_t oldpathc; | |
486 | Char *bufnext, c, patbuf[MAXPATHLEN]; | |
224c7076 A |
487 | |
488 | qpatnext = globtilde(pattern, patbuf, MAXPATHLEN, pglob); | |
489 | oldpathc = pglob->gl_pathc; | |
490 | bufnext = patbuf; | |
491 | ||
492 | /* We don't need to check for buffer overflow any more. */ | |
493 | while ((c = *qpatnext++) != EOS) { | |
494 | if (c & M_PROTECT) { | |
495 | *bufnext++ = CHAR(c); | |
496 | continue; | |
497 | } /* else */ | |
498 | switch (c) { | |
499 | case LBRACKET: | |
500 | c = *qpatnext; | |
501 | if (c == NOT) | |
502 | ++qpatnext; | |
503 | if (*qpatnext == EOS || | |
fbd86d4c | 504 | g_strchr(qpatnext+1, RBRACKET) == NULL) { |
224c7076 A |
505 | *bufnext++ = LBRACKET; |
506 | if (c == NOT) | |
507 | --qpatnext; | |
508 | break; | |
509 | } | |
510 | *bufnext++ = M_SET; | |
511 | if (c == NOT) | |
512 | *bufnext++ = M_NOT; | |
513 | c = *qpatnext++; | |
514 | do { | |
515 | *bufnext++ = CHAR(c); | |
516 | if (*qpatnext == RANGE && | |
517 | (c = qpatnext[1]) != RBRACKET) { | |
518 | *bufnext++ = M_RNG; | |
519 | *bufnext++ = CHAR(c); | |
520 | qpatnext += 2; | |
521 | } | |
522 | } while ((c = *qpatnext++) != RBRACKET); | |
523 | pglob->gl_flags |= GLOB_MAGCHAR; | |
524 | *bufnext++ = M_END; | |
525 | break; | |
526 | case QUESTION: | |
527 | pglob->gl_flags |= GLOB_MAGCHAR; | |
528 | *bufnext++ = M_ONE; | |
529 | break; | |
530 | case STAR: | |
531 | pglob->gl_flags |= GLOB_MAGCHAR; | |
532 | /* collapse adjacent stars to one, | |
533 | * to avoid exponential behavior | |
534 | */ | |
535 | if (bufnext == patbuf || bufnext[-1] != M_ALL) | |
536 | *bufnext++ = M_ALL; | |
537 | break; | |
538 | default: | |
539 | *bufnext++ = CHAR(c); | |
540 | break; | |
541 | } | |
542 | } | |
543 | *bufnext = EOS; | |
544 | #ifdef DEBUG | |
545 | qprintf("glob0:", patbuf); | |
546 | #endif | |
547 | ||
548 | if ((err = glob1(patbuf, pglob, limit, loc)) != 0) | |
549 | return(err); | |
550 | ||
551 | /* | |
552 | * If there was no match we are going to append the pattern | |
553 | * if GLOB_NOCHECK was specified or if GLOB_NOMAGIC was specified | |
554 | * and the pattern did not contain any magic characters | |
555 | * GLOB_NOMAGIC is there just for compatibility with csh. | |
556 | */ | |
557 | if (pglob->gl_pathc == oldpathc) { | |
558 | if (((pglob->gl_flags & GLOB_NOCHECK) || | |
559 | ((pglob->gl_flags & GLOB_NOMAGIC) && | |
560 | !(pglob->gl_flags & GLOB_MAGCHAR)))) | |
561 | return(globextend(pattern, pglob, limit, loc)); | |
562 | else | |
563 | return(GLOB_NOMATCH); | |
564 | } | |
565 | if (!(pglob->gl_flags & GLOB_NOSORT)) | |
566 | qsort(pglob->gl_pathv + pglob->gl_offs + oldpathc, | |
567 | pglob->gl_pathc - oldpathc, sizeof(char *), compare); | |
568 | return(0); | |
569 | } | |
570 | ||
b5d655f7 A |
571 | #ifndef BUILDING_VARIANT |
572 | __private_extern__ int | |
fbd86d4c | 573 | compare(const void *p, const void *q) |
224c7076 A |
574 | { |
575 | return(strcoll(*(char **)p, *(char **)q)); | |
576 | } | |
b5d655f7 | 577 | #endif /* BUILDING_VARIANT */ |
224c7076 A |
578 | |
579 | static int | |
fbd86d4c | 580 | glob1(Char *pattern, glob_t *pglob, struct glob_limit *limit, locale_t loc) |
224c7076 A |
581 | { |
582 | Char pathbuf[MAXPATHLEN]; | |
583 | ||
584 | /* A null pathname is invalid -- POSIX 1003.1 sect. 2.4. */ | |
585 | if (*pattern == EOS) | |
586 | return(0); | |
587 | return(glob2(pathbuf, pathbuf, pathbuf + MAXPATHLEN - 1, | |
588 | pattern, pglob, limit, loc)); | |
589 | } | |
224c7076 A |
590 | |
591 | /* | |
592 | * The functions glob2 and glob3 are mutually recursive; there is one level | |
593 | * of recursion for each segment in the pattern that contains one or more | |
594 | * meta characters. | |
595 | */ | |
b5d655f7 | 596 | static int |
fbd86d4c A |
597 | glob2(Char *pathbuf, Char *pathend, Char *pathend_last, Char *pattern, |
598 | glob_t *pglob, struct glob_limit *limit, locale_t loc) | |
224c7076 A |
599 | { |
600 | struct stat sb; | |
601 | Char *p, *q; | |
602 | int anymeta; | |
603 | ||
604 | /* | |
605 | * Loop over pattern segments until end of pattern or until | |
606 | * segment with meta character found. | |
607 | */ | |
608 | for (anymeta = 0;;) { | |
609 | if (*pattern == EOS) { /* End of pattern? */ | |
610 | *pathend = EOS; | |
611 | if (g_lstat(pathbuf, &sb, pglob, loc)) | |
612 | return(0); | |
613 | ||
fbd86d4c A |
614 | if ((pglob->gl_flags & GLOB_LIMIT) && |
615 | limit->l_stat++ >= GLOB_LIMIT_STAT) { | |
616 | errno = 0; | |
617 | *pathend++ = SEP; | |
618 | *pathend = EOS; | |
619 | return GLOB_NOSPACE; | |
620 | } | |
224c7076 A |
621 | if (((pglob->gl_flags & GLOB_MARK) && |
622 | pathend[-1] != SEP) && (S_ISDIR(sb.st_mode) | |
623 | || (S_ISLNK(sb.st_mode) && | |
624 | (g_stat(pathbuf, &sb, pglob, loc) == 0) && | |
625 | S_ISDIR(sb.st_mode)))) { | |
626 | if (pathend + 1 > pathend_last) | |
627 | return (GLOB_ABORTED); | |
628 | *pathend++ = SEP; | |
629 | *pathend = EOS; | |
630 | } | |
631 | ++pglob->gl_matchc; | |
632 | return(globextend(pathbuf, pglob, limit, loc)); | |
633 | } | |
634 | ||
635 | /* Find end of next segment, copy tentatively to pathend. */ | |
636 | q = pathend; | |
637 | p = pattern; | |
638 | while (*p != EOS && *p != SEP) { | |
639 | if (ismeta(*p)) | |
640 | anymeta = 1; | |
641 | if (q + 1 > pathend_last) | |
642 | return (GLOB_ABORTED); | |
643 | *q++ = *p++; | |
644 | } | |
645 | ||
646 | if (!anymeta) { /* No expansion, do next segment. */ | |
647 | pathend = q; | |
648 | pattern = p; | |
649 | while (*pattern == SEP) { | |
650 | if (pathend + 1 > pathend_last) | |
651 | return (GLOB_ABORTED); | |
652 | *pathend++ = *pattern++; | |
653 | } | |
654 | } else /* Need expansion, recurse. */ | |
655 | return(glob3(pathbuf, pathend, pathend_last, pattern, p, | |
656 | pglob, limit, loc)); | |
657 | } | |
658 | /* NOTREACHED */ | |
659 | } | |
660 | ||
b5d655f7 | 661 | static int |
fbd86d4c A |
662 | glob3(Char *pathbuf, Char *pathend, Char *pathend_last, |
663 | Char *pattern, Char *restpattern, | |
664 | glob_t *pglob, struct glob_limit *limit, locale_t loc) | |
224c7076 A |
665 | { |
666 | struct dirent *dp; | |
667 | DIR *dirp; | |
668 | int err; | |
669 | char buf[MAXPATHLEN]; | |
670 | ||
671 | /* | |
672 | * The readdirfunc declaration can't be prototyped, because it is | |
673 | * assigned, below, to two functions which are prototyped in glob.h | |
674 | * and dirent.h as taking pointers to differently typed opaque | |
675 | * structures. | |
676 | */ | |
677 | struct dirent *(*readdirfunc)(); | |
678 | ||
679 | if (pathend > pathend_last) | |
680 | return (GLOB_ABORTED); | |
681 | *pathend = EOS; | |
682 | errno = 0; | |
683 | ||
684 | if ((dirp = g_opendir(pathbuf, pglob, loc)) == NULL) { | |
685 | /* TODO: don't call for ENOENT or ENOTDIR? */ | |
686 | if (pglob->gl_errfunc) { | |
687 | if (g_Ctoc(pathbuf, buf, sizeof(buf), loc)) | |
688 | return (GLOB_ABORTED); | |
34e8f829 A |
689 | #ifdef __BLOCKS__ |
690 | if (pglob->gl_flags & _GLOB_ERR_BLOCK) { | |
691 | if (pglob->gl_errblk(buf, errno)) | |
692 | return (GLOB_ABORTED); | |
693 | } else | |
694 | #endif /* __BLOCKS__ */ | |
224c7076 A |
695 | if (pglob->gl_errfunc(buf, errno)) |
696 | return (GLOB_ABORTED); | |
697 | } | |
698 | if (pglob->gl_flags & GLOB_ERR) | |
699 | return (GLOB_ABORTED); | |
700 | return(0); | |
701 | } | |
702 | ||
703 | err = 0; | |
704 | ||
705 | /* Search directory for matching names. */ | |
706 | if (pglob->gl_flags & GLOB_ALTDIRFUNC) | |
707 | readdirfunc = pglob->gl_readdir; | |
708 | else | |
709 | readdirfunc = readdir; | |
710 | while ((dp = (*readdirfunc)(dirp))) { | |
fbd86d4c | 711 | char *sc; |
224c7076 A |
712 | Char *dc; |
713 | wchar_t wc; | |
714 | size_t clen; | |
715 | mbstate_t mbs; | |
716 | ||
fbd86d4c A |
717 | if ((pglob->gl_flags & GLOB_LIMIT) && |
718 | limit->l_readdir++ >= GLOB_LIMIT_READDIR) { | |
719 | errno = 0; | |
720 | *pathend++ = SEP; | |
721 | *pathend = EOS; | |
722 | return GLOB_NOSPACE; | |
723 | } | |
724 | ||
224c7076 A |
725 | /* Initial DOT must be matched literally. */ |
726 | if (dp->d_name[0] == DOT && *pattern != DOT) | |
727 | continue; | |
728 | memset(&mbs, 0, sizeof(mbs)); | |
729 | dc = pathend; | |
fbd86d4c | 730 | sc = dp->d_name; |
224c7076 | 731 | while (dc < pathend_last) { |
fbd86d4c | 732 | clen = mbrtowc_l(&wc, sc, MB_LEN_MAX, &mbs, loc); |
224c7076 A |
733 | if (clen == (size_t)-1 || clen == (size_t)-2) { |
734 | wc = *sc; | |
735 | clen = 1; | |
736 | memset(&mbs, 0, sizeof(mbs)); | |
737 | } | |
738 | if ((*dc++ = wc) == EOS) | |
739 | break; | |
740 | sc += clen; | |
741 | } | |
742 | if (!match(pathend, pattern, restpattern, loc)) { | |
743 | *pathend = EOS; | |
744 | continue; | |
745 | } | |
746 | err = glob2(pathbuf, --dc, pathend_last, restpattern, | |
747 | pglob, limit, loc); | |
748 | if (err) | |
749 | break; | |
750 | } | |
751 | ||
752 | if (pglob->gl_flags & GLOB_ALTDIRFUNC) | |
753 | (*pglob->gl_closedir)(dirp); | |
754 | else | |
755 | closedir(dirp); | |
756 | return(err); | |
757 | } | |
758 | ||
759 | ||
b5d655f7 | 760 | #ifndef BUILDING_VARIANT |
224c7076 A |
761 | /* |
762 | * Extend the gl_pathv member of a glob_t structure to accomodate a new item, | |
763 | * add the new item, and update gl_pathc. | |
764 | * | |
765 | * This assumes the BSD realloc, which only copies the block when its size | |
766 | * crosses a power-of-two boundary; for v7 realloc, this would cause quadratic | |
767 | * behavior. | |
768 | * | |
769 | * Return 0 if new item added, error code if memory couldn't be allocated. | |
770 | * | |
771 | * Invariant of the glob_t structure: | |
772 | * Either gl_pathc is zero and gl_pathv is NULL; or gl_pathc > 0 and | |
773 | * gl_pathv points to (gl_offs + gl_pathc + 1) items. | |
774 | */ | |
775 | __private_extern__ int | |
fbd86d4c | 776 | globextend(const Char *path, glob_t *pglob, struct glob_limit *limit, locale_t loc) |
224c7076 A |
777 | { |
778 | char **pathv; | |
fbd86d4c | 779 | size_t i, newsize, len; |
224c7076 A |
780 | char *copy; |
781 | const Char *p; | |
782 | ||
224c7076 | 783 | newsize = sizeof(*pathv) * (2 + pglob->gl_pathc + pglob->gl_offs); |
fbd86d4c A |
784 | if ((pglob->gl_flags & GLOB_LIMIT) && |
785 | newsize > GLOB_LIMIT_PATH * sizeof(*pathv)) | |
786 | goto nospace; | |
224c7076 A |
787 | pathv = pglob->gl_pathv ? |
788 | realloc((char *)pglob->gl_pathv, newsize) : | |
789 | malloc(newsize); | |
790 | if (pathv == NULL) { | |
791 | if (pglob->gl_pathv) { | |
792 | free(pglob->gl_pathv); | |
793 | pglob->gl_pathv = NULL; | |
794 | } | |
795 | return(GLOB_NOSPACE); | |
796 | } | |
797 | ||
798 | if (pglob->gl_pathv == NULL && pglob->gl_offs > 0) { | |
799 | /* first time around -- clear initial gl_offs items */ | |
800 | pathv += pglob->gl_offs; | |
fbd86d4c | 801 | for (i = pglob->gl_offs + 1; --i > 0; ) |
224c7076 A |
802 | *--pathv = NULL; |
803 | } | |
804 | pglob->gl_pathv = pathv; | |
805 | ||
806 | for (p = path; *p++;) | |
807 | continue; | |
808 | len = MB_CUR_MAX_L(loc) * (size_t)(p - path); /* XXX overallocation */ | |
fbd86d4c | 809 | limit->l_string += len; |
224c7076 A |
810 | if ((copy = malloc(len)) != NULL) { |
811 | if (g_Ctoc(path, copy, len, loc)) { | |
812 | free(copy); | |
813 | return (GLOB_NOSPACE); | |
814 | } | |
815 | pathv[pglob->gl_offs + pglob->gl_pathc++] = copy; | |
816 | } | |
817 | pathv[pglob->gl_offs + pglob->gl_pathc] = NULL; | |
fbd86d4c A |
818 | |
819 | if ((pglob->gl_flags & GLOB_LIMIT) && | |
820 | (newsize + limit->l_string) >= GLOB_LIMIT_STRING) | |
821 | goto nospace; | |
822 | ||
224c7076 | 823 | return(copy == NULL ? GLOB_NOSPACE : 0); |
fbd86d4c A |
824 | nospace: |
825 | errno = 0; | |
826 | return GLOB_NOSPACE; | |
224c7076 A |
827 | } |
828 | ||
829 | /* | |
830 | * pattern matching function for filenames. Each occurrence of the * | |
831 | * pattern causes a recursion level. | |
832 | */ | |
b5d655f7 | 833 | __private_extern__ int |
fbd86d4c | 834 | match(Char *name, Char *pat, Char *patend, locale_t loc) |
224c7076 A |
835 | { |
836 | int ok, negate_range; | |
837 | Char c, k; | |
838 | ||
839 | while (pat < patend) { | |
840 | c = *pat++; | |
841 | switch (c & M_MASK) { | |
842 | case M_ALL: | |
843 | if (pat == patend) | |
844 | return(1); | |
845 | do | |
846 | if (match(name, pat, patend, loc)) | |
847 | return(1); | |
848 | while (*name++ != EOS); | |
849 | return(0); | |
850 | case M_ONE: | |
851 | if (*name++ == EOS) | |
852 | return(0); | |
853 | break; | |
854 | case M_SET: | |
855 | ok = 0; | |
856 | if ((k = *name++) == EOS) | |
857 | return(0); | |
858 | if ((negate_range = ((*pat & M_MASK) == M_NOT)) != EOS) | |
859 | ++pat; | |
860 | while (((c = *pat++) & M_MASK) != M_END) | |
861 | if ((*pat & M_MASK) == M_RNG) { | |
862 | if (loc->__collate_load_error ? | |
863 | CHAR(c) <= CHAR(k) && CHAR(k) <= CHAR(pat[1]) : | |
864 | __collate_range_cmp(CHAR(c), CHAR(k), loc) <= 0 | |
865 | && __collate_range_cmp(CHAR(k), CHAR(pat[1]), loc) <= 0 | |
866 | ) | |
867 | ok = 1; | |
868 | pat += 2; | |
869 | } else if (c == k) | |
870 | ok = 1; | |
871 | if (ok == negate_range) | |
872 | return(0); | |
873 | break; | |
874 | default: | |
875 | if (*name++ != c) | |
876 | return(0); | |
877 | break; | |
878 | } | |
879 | } | |
880 | return(*name == EOS); | |
881 | } | |
882 | ||
883 | /* Free allocated data belonging to a glob_t structure. */ | |
884 | void | |
fbd86d4c | 885 | globfree(glob_t *pglob) |
224c7076 | 886 | { |
fbd86d4c | 887 | size_t i; |
224c7076 A |
888 | char **pp; |
889 | ||
890 | if (pglob->gl_pathv != NULL) { | |
891 | pp = pglob->gl_pathv + pglob->gl_offs; | |
892 | for (i = pglob->gl_pathc; i--; ++pp) | |
893 | if (*pp) | |
894 | free(*pp); | |
895 | free(pglob->gl_pathv); | |
896 | pglob->gl_pathv = NULL; | |
897 | } | |
898 | } | |
b5d655f7 | 899 | #endif /* !BUILDING_VARIANT */ |
224c7076 A |
900 | |
901 | static DIR * | |
fbd86d4c | 902 | g_opendir(Char *str, glob_t *pglob, locale_t loc) |
224c7076 A |
903 | { |
904 | char buf[MAXPATHLEN]; | |
905 | ||
906 | if (!*str) | |
907 | strcpy(buf, "."); | |
908 | else { | |
909 | if (g_Ctoc(str, buf, sizeof(buf), loc)) | |
910 | return (NULL); | |
911 | } | |
912 | ||
913 | if (pglob->gl_flags & GLOB_ALTDIRFUNC) | |
914 | return((*pglob->gl_opendir)(buf)); | |
915 | ||
916 | return(opendir(buf)); | |
917 | } | |
224c7076 A |
918 | |
919 | static int | |
fbd86d4c | 920 | g_lstat(Char *fn, struct stat *sb, glob_t *pglob, locale_t loc) |
224c7076 A |
921 | { |
922 | char buf[MAXPATHLEN]; | |
923 | ||
924 | if (g_Ctoc(fn, buf, sizeof(buf), loc)) { | |
925 | errno = ENAMETOOLONG; | |
926 | return (-1); | |
927 | } | |
928 | if (pglob->gl_flags & GLOB_ALTDIRFUNC) | |
929 | return((*pglob->gl_lstat)(buf, sb)); | |
930 | return(lstat(buf, sb)); | |
931 | } | |
932 | ||
933 | static int | |
fbd86d4c | 934 | g_stat(Char *fn, struct stat *sb, glob_t *pglob, locale_t loc) |
224c7076 A |
935 | { |
936 | char buf[MAXPATHLEN]; | |
937 | ||
938 | if (g_Ctoc(fn, buf, sizeof(buf), loc)) { | |
939 | errno = ENAMETOOLONG; | |
940 | return (-1); | |
941 | } | |
942 | if (pglob->gl_flags & GLOB_ALTDIRFUNC) | |
943 | return((*pglob->gl_stat)(buf, sb)); | |
944 | return(stat(buf, sb)); | |
945 | } | |
946 | ||
947 | #ifndef BUILDING_VARIANT | |
fbd86d4c A |
948 | __private_extern__ const Char * |
949 | g_strchr(const Char *str, wchar_t ch) | |
224c7076 | 950 | { |
fbd86d4c | 951 | |
224c7076 A |
952 | do { |
953 | if (*str == ch) | |
954 | return (str); | |
955 | } while (*str++); | |
956 | return (NULL); | |
957 | } | |
958 | ||
959 | __private_extern__ int | |
fbd86d4c | 960 | g_Ctoc(const Char *str, char *buf, size_t len, locale_t loc) |
224c7076 A |
961 | { |
962 | mbstate_t mbs; | |
963 | size_t clen; | |
964 | int mb_cur_max = MB_CUR_MAX_L(loc); | |
965 | ||
966 | memset(&mbs, 0, sizeof(mbs)); | |
967 | while (len >= mb_cur_max) { | |
968 | clen = wcrtomb_l(buf, *str, &mbs, loc); | |
969 | if (clen == (size_t)-1) | |
970 | return (1); | |
971 | if (*str == L'\0') | |
972 | return (0); | |
973 | str++; | |
974 | buf += clen; | |
975 | len -= clen; | |
976 | } | |
977 | return (1); | |
978 | } | |
979 | ||
980 | #ifdef DEBUG | |
981 | static void | |
fbd86d4c | 982 | qprintf(const char *str, Char *s) |
224c7076 A |
983 | { |
984 | Char *p; | |
985 | ||
986 | (void)printf("%s:\n", str); | |
987 | for (p = s; *p; p++) | |
988 | (void)printf("%c", CHAR(*p)); | |
989 | (void)printf("\n"); | |
990 | for (p = s; *p; p++) | |
991 | (void)printf("%c", *p & M_PROTECT ? '"' : ' '); | |
992 | (void)printf("\n"); | |
993 | for (p = s; *p; p++) | |
994 | (void)printf("%c", ismeta(*p) ? '_' : ' '); | |
995 | (void)printf("\n"); | |
996 | } | |
997 | #endif | |
998 | #endif /* !BUILDING_VARIANT */ |