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. 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 * @(#)glob.c 8.3 (Berkeley) 10/13/93 33 * $FreeBSD: src/lib/libc/gen/glob.c,v 1.27 2008/06/26 07:12:35 mtm Exp $ 34 * $DragonFly: src/lib/libc/gen/glob.c,v 1.6 2005/12/07 02:28:15 corecode Exp $ 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 /* 63 * Some notes on multibyte character support: 64 * 1. Patterns with illegal byte sequences match nothing - even if 65 * GLOB_NOCHECK is specified. 66 * 2. Illegal byte sequences in filenames are handled by treating them as 67 * single-byte characters with a value of the first byte of the sequence 68 * cast to wchar_t. 69 * 3. State-dependent encodings are not currently supported. 70 */ 71 72 #include <sys/param.h> 73 #include <sys/stat.h> 74 75 #include <ctype.h> 76 #include <dirent.h> 77 #include <errno.h> 78 #include <glob.h> 79 #include <limits.h> 80 #include <pwd.h> 81 #include <stdint.h> 82 #include <stdio.h> 83 #include <stdlib.h> 84 #include <string.h> 85 #include <unistd.h> 86 #include <wchar.h> 87 88 #include "collate.h" 89 90 #define DOLLAR '$' 91 #define DOT '.' 92 #define EOS '\0' 93 #define LBRACKET '[' 94 #define NOT '!' 95 #define QUESTION '?' 96 #define QUOTE '\\' 97 #define RANGE '-' 98 #define RBRACKET ']' 99 #define SEP '/' 100 #define STAR '*' 101 #define TILDE '~' 102 #define UNDERSCORE '_' 103 #define LBRACE '{' 104 #define RBRACE '}' 105 #define SLASH '/' 106 #define COMMA ',' 107 108 #ifndef DEBUG 109 110 #define M_QUOTE 0x8000000000ULL 111 #define M_PROTECT 0x4000000000ULL 112 #define M_MASK 0xffffffffffULL 113 #define M_CHAR 0x00ffffffffULL 114 115 typedef uint_fast64_t Char; 116 117 #else 118 119 #define M_QUOTE 0x80 120 #define M_PROTECT 0x40 121 #define M_MASK 0xff 122 #define M_CHAR 0x7f 123 124 typedef char Char; 125 126 #endif 127 128 129 #define CHAR(c) ((Char)((c)&M_CHAR)) 130 #define META(c) ((Char)((c)|M_QUOTE)) 131 #define M_ALL META('*') 132 #define M_END META(']') 133 #define M_NOT META('!') 134 #define M_ONE META('?') 135 #define M_RNG META('-') 136 #define M_SET META('[') 137 #define ismeta(c) (((c)&M_QUOTE) != 0) 138 139 140 #define GLOB_LIMIT_MALLOC 65536 141 #define GLOB_LIMIT_STAT 128 142 #define GLOB_LIMIT_READDIR 16384 143 144 #define GLOB_INDEX_MALLOC 0 145 #define GLOB_INDEX_STAT 1 146 #define GLOB_INDEX_READDIR 2 147 148 static int compare(const void *, const void *); 149 static int g_Ctoc(const Char *, char *, size_t); 150 static int g_lstat(const Char *, struct stat *, glob_t *); 151 static DIR *g_opendir(const Char *, glob_t *); 152 static const Char *g_strchr(const Char *, wchar_t); 153 #ifdef notdef 154 static Char *g_strcat(Char *, const Char *); 155 #endif 156 static int g_stat(const Char *, struct stat *, glob_t *); 157 static int glob0(const Char *, glob_t *, size_t *); 158 static int glob1(const Char *, glob_t *, size_t *); 159 static int glob2(Char *, Char *, Char *, const Char *, glob_t *, 160 size_t *); 161 static int glob3(Char *, Char *, Char *, const Char *, const Char *, 162 glob_t *, size_t *); 163 static int globextend(const Char *, glob_t *, size_t *); 164 static const Char * 165 globtilde(const Char *, Char *, size_t, glob_t *); 166 static int globexp1(const Char *, glob_t *, size_t *); 167 static int globexp2(const Char *, const Char *, glob_t *, int *, size_t *); 168 static int match(const Char *, const Char *, const Char *); 169 #ifdef DEBUG 170 static void qprintf(const char *, const Char *); 171 #endif 172 173 /* 0 = malloc(), 1 = stat(), 2 = readdir() */ 174 static size_t limits[] = { 0, 0, 0 }; 175 176 int 177 glob(const char *pattern, int flags, int (*errfunc)(const char *, int), 178 glob_t *pglob) 179 { 180 const char *patnext; 181 size_t limit; 182 Char *bufnext, *bufend, patbuf[MAXPATHLEN], prot; 183 mbstate_t mbs; 184 wchar_t wc; 185 size_t clen; 186 187 patnext = pattern; 188 if (!(flags & GLOB_APPEND)) { 189 pglob->gl_pathc = 0; 190 pglob->gl_pathv = NULL; 191 if (!(flags & GLOB_DOOFFS)) 192 pglob->gl_offs = 0; 193 } 194 if (flags & GLOB_LIMIT) { 195 limit = pglob->gl_matchc; 196 if (limit == 0) 197 limit = ARG_MAX; 198 } else 199 limit = 0; 200 pglob->gl_flags = flags & ~GLOB_MAGCHAR; 201 pglob->gl_errfunc = errfunc; 202 pglob->gl_matchc = 0; 203 204 bufnext = patbuf; 205 bufend = bufnext + MAXPATHLEN - 1; 206 if (flags & GLOB_NOESCAPE) { 207 memset(&mbs, 0, sizeof(mbs)); 208 while (bufend - bufnext >= MB_CUR_MAX) { 209 clen = mbrtowc(&wc, patnext, MB_LEN_MAX, &mbs); 210 if (clen == (size_t)-1 || clen == (size_t)-2) 211 return (GLOB_NOMATCH); 212 else if (clen == 0) 213 break; 214 *bufnext++ = wc; 215 patnext += clen; 216 } 217 } else { 218 /* Protect the quoted characters. */ 219 memset(&mbs, 0, sizeof(mbs)); 220 while (bufend - bufnext >= MB_CUR_MAX) { 221 if (*patnext == QUOTE) { 222 if (*++patnext == EOS) { 223 *bufnext++ = QUOTE | M_PROTECT; 224 continue; 225 } 226 prot = M_PROTECT; 227 } else 228 prot = 0; 229 clen = mbrtowc(&wc, patnext, MB_LEN_MAX, &mbs); 230 if (clen == (size_t)-1 || clen == (size_t)-2) 231 return (GLOB_NOMATCH); 232 else if (clen == 0) 233 break; 234 *bufnext++ = wc | prot; 235 patnext += clen; 236 } 237 } 238 *bufnext = EOS; 239 240 if (flags & GLOB_BRACE) 241 return globexp1(patbuf, pglob, &limit); 242 else 243 return glob0(patbuf, pglob, &limit); 244 } 245 246 /* 247 * Expand recursively a glob {} pattern. When there is no more expansion 248 * invoke the standard globbing routine to glob the rest of the magic 249 * characters 250 */ 251 static int 252 globexp1(const Char *pattern, glob_t *pglob, size_t *limit) 253 { 254 const Char *ptr = pattern; 255 int rv; 256 257 /* Protect a single {}, for find(1), like csh */ 258 if (pattern[0] == LBRACE && pattern[1] == RBRACE && pattern[2] == EOS) 259 return glob0(pattern, pglob, limit); 260 261 while ((ptr = g_strchr(ptr, LBRACE)) != NULL) 262 if (!globexp2(ptr, pattern, pglob, &rv, limit)) 263 return rv; 264 265 return glob0(pattern, pglob, limit); 266 } 267 268 269 /* 270 * Recursive brace globbing helper. Tries to expand a single brace. 271 * If it succeeds then it invokes globexp1 with the new pattern. 272 * If it fails then it tries to glob the rest of the pattern and returns. 273 */ 274 static int 275 globexp2(const Char *ptr, const Char *pattern, glob_t *pglob, int *rv, 276 size_t *limit) 277 { 278 int i; 279 Char *lm, *ls; 280 const Char *pe, *pm, *pm1, *pl; 281 Char patbuf[MAXPATHLEN]; 282 283 /* copy part up to the brace */ 284 for (lm = patbuf, pm = pattern; pm != ptr; *lm++ = *pm++) 285 continue; 286 *lm = EOS; 287 ls = lm; 288 289 /* Find the balanced brace */ 290 for (i = 0, pe = ++ptr; *pe; pe++) 291 if (*pe == LBRACKET) { 292 /* Ignore everything between [] */ 293 for (pm = pe++; *pe != RBRACKET && *pe != EOS; pe++) 294 continue; 295 if (*pe == EOS) { 296 /* 297 * We could not find a matching RBRACKET. 298 * Ignore and just look for RBRACE 299 */ 300 pe = pm; 301 } 302 } 303 else if (*pe == LBRACE) 304 i++; 305 else if (*pe == RBRACE) { 306 if (i == 0) 307 break; 308 i--; 309 } 310 311 /* Non matching braces; just glob the pattern */ 312 if (i != 0 || *pe == EOS) { 313 *rv = glob0(patbuf, pglob, limit); 314 return 0; 315 } 316 317 for (i = 0, pl = pm = ptr; pm <= pe; pm++) 318 switch (*pm) { 319 case LBRACKET: 320 /* Ignore everything between [] */ 321 for (pm1 = pm++; *pm != RBRACKET && *pm != EOS; pm++) 322 continue; 323 if (*pm == EOS) { 324 /* 325 * We could not find a matching RBRACKET. 326 * Ignore and just look for RBRACE 327 */ 328 pm = pm1; 329 } 330 break; 331 332 case LBRACE: 333 i++; 334 break; 335 336 case RBRACE: 337 if (i) { 338 i--; 339 break; 340 } 341 /* FALLTHROUGH */ 342 case COMMA: 343 if (i && *pm == COMMA) 344 break; 345 else { 346 /* Append the current string */ 347 for (lm = ls; (pl < pm); *lm++ = *pl++) 348 continue; 349 /* 350 * Append the rest of the pattern after the 351 * closing brace 352 */ 353 for (pl = pe + 1; (*lm++ = *pl++) != EOS;) 354 continue; 355 356 /* Expand the current pattern */ 357 #ifdef DEBUG 358 qprintf("globexp2:", patbuf); 359 #endif 360 *rv = globexp1(patbuf, pglob, limit); 361 362 /* move after the comma, to the next string */ 363 pl = pm + 1; 364 } 365 break; 366 367 default: 368 break; 369 } 370 *rv = 0; 371 return 0; 372 } 373 374 375 376 /* 377 * expand tilde from the passwd file. 378 */ 379 static const Char * 380 globtilde(const Char *pattern, Char *patbuf, size_t patbuf_len, glob_t *pglob) 381 { 382 struct passwd *pwd; 383 char *h; 384 const Char *p; 385 Char *b, *eb; 386 387 if (*pattern != TILDE || !(pglob->gl_flags & GLOB_TILDE)) 388 return pattern; 389 390 /* 391 * Copy up to the end of the string or / 392 */ 393 eb = &patbuf[patbuf_len - 1]; 394 for (p = pattern + 1, h = (char *) patbuf; 395 h < (char *)eb && *p && *p != SLASH; *h++ = *p++) 396 continue; 397 398 *h = EOS; 399 400 if (((char *) patbuf)[0] == EOS) { 401 /* 402 * handle a plain ~ or ~/ by expanding $HOME first (iff 403 * we're not running setuid or setgid) and then trying 404 * the password file 405 */ 406 if (issetugid() != 0 || 407 (h = getenv("HOME")) == NULL) { 408 if (((h = getlogin()) != NULL && 409 (pwd = getpwnam(h)) != NULL) || 410 (pwd = getpwuid(getuid())) != NULL) 411 h = pwd->pw_dir; 412 else 413 return pattern; 414 } 415 } 416 else { 417 /* 418 * Expand a ~user 419 */ 420 if ((pwd = getpwnam((char*) patbuf)) == NULL) 421 return pattern; 422 else 423 h = pwd->pw_dir; 424 } 425 426 /* Copy the home directory */ 427 for (b = patbuf; b < eb && *h; *b++ = *h++) 428 continue; 429 430 /* Append the rest of the pattern */ 431 while (b < eb && (*b++ = *p++) != EOS) 432 continue; 433 *b = EOS; 434 435 return patbuf; 436 } 437 438 439 /* 440 * The main glob() routine: compiles the pattern (optionally processing 441 * quotes), calls glob1() to do the real pattern matching, and finally 442 * sorts the list (unless unsorted operation is requested). Returns 0 443 * if things went well, nonzero if errors occurred. 444 */ 445 static int 446 glob0(const Char *pattern, glob_t *pglob, size_t *limit) 447 { 448 const Char *qpatnext; 449 int c, err; 450 size_t oldpathc; 451 Char *bufnext, patbuf[MAXPATHLEN]; 452 453 qpatnext = globtilde(pattern, patbuf, MAXPATHLEN, pglob); 454 oldpathc = pglob->gl_pathc; 455 bufnext = patbuf; 456 457 /* We don't need to check for buffer overflow any more. */ 458 while ((c = *qpatnext++) != EOS) { 459 switch (c) { 460 case LBRACKET: 461 c = *qpatnext; 462 if (c == NOT) 463 ++qpatnext; 464 if (*qpatnext == EOS || 465 g_strchr(qpatnext+1, RBRACKET) == NULL) { 466 *bufnext++ = LBRACKET; 467 if (c == NOT) 468 --qpatnext; 469 break; 470 } 471 *bufnext++ = M_SET; 472 if (c == NOT) 473 *bufnext++ = M_NOT; 474 c = *qpatnext++; 475 do { 476 *bufnext++ = CHAR(c); 477 if (*qpatnext == RANGE && 478 (c = qpatnext[1]) != RBRACKET) { 479 *bufnext++ = M_RNG; 480 *bufnext++ = CHAR(c); 481 qpatnext += 2; 482 } 483 } while ((c = *qpatnext++) != RBRACKET); 484 pglob->gl_flags |= GLOB_MAGCHAR; 485 *bufnext++ = M_END; 486 break; 487 case QUESTION: 488 pglob->gl_flags |= GLOB_MAGCHAR; 489 *bufnext++ = M_ONE; 490 break; 491 case STAR: 492 pglob->gl_flags |= GLOB_MAGCHAR; 493 /* collapse adjacent stars to one, 494 * to avoid exponential behavior 495 */ 496 if (bufnext == patbuf || bufnext[-1] != M_ALL) 497 *bufnext++ = M_ALL; 498 break; 499 default: 500 *bufnext++ = CHAR(c); 501 break; 502 } 503 } 504 *bufnext = EOS; 505 #ifdef DEBUG 506 qprintf("glob0:", patbuf); 507 #endif 508 509 if ((err = glob1(patbuf, pglob, limit)) != 0) 510 return(err); 511 512 /* 513 * If there was no match we are going to append the pattern 514 * if GLOB_NOCHECK was specified or if GLOB_NOMAGIC was specified 515 * and the pattern did not contain any magic characters 516 * GLOB_NOMAGIC is there just for compatibility with csh. 517 */ 518 if (pglob->gl_pathc == oldpathc) { 519 if (((pglob->gl_flags & GLOB_NOCHECK) || 520 ((pglob->gl_flags & GLOB_NOMAGIC) && 521 !(pglob->gl_flags & GLOB_MAGCHAR)))) 522 return(globextend(pattern, pglob, limit)); 523 else 524 return(GLOB_NOMATCH); 525 } 526 if (!(pglob->gl_flags & GLOB_NOSORT)) 527 qsort(pglob->gl_pathv + pglob->gl_offs + oldpathc, 528 pglob->gl_pathc - oldpathc, sizeof(char *), compare); 529 return(0); 530 } 531 532 static int 533 compare(const void *p, const void *q) 534 { 535 return(strcmp(*(const char * const *)p, *(const char * const *)q)); 536 } 537 538 static int 539 glob1(const Char *pattern, glob_t *pglob, size_t *limit) 540 { 541 Char pathbuf[MAXPATHLEN]; 542 543 /* A null pathname is invalid -- POSIX 1003.1 sect. 2.4. */ 544 if (*pattern == EOS) 545 return(0); 546 return(glob2(pathbuf, pathbuf, pathbuf + MAXPATHLEN - 1, 547 pattern, pglob, limit)); 548 } 549 550 /* 551 * The functions glob2 and glob3 are mutually recursive; there is one level 552 * of recursion for each segment in the pattern that contains one or more 553 * meta characters. 554 */ 555 static int 556 glob2(Char *pathbuf, Char *pathend, Char *pathend_last, const Char *pattern, 557 glob_t *pglob, size_t *limit) 558 { 559 struct stat sb; 560 const Char *p; 561 Char *q; 562 int anymeta; 563 564 /* 565 * Loop over pattern segments until end of pattern or until 566 * segment with meta character found. 567 */ 568 for (anymeta = 0;;) { 569 if (*pattern == EOS) { /* End of pattern? */ 570 *pathend = EOS; 571 if (g_lstat(pathbuf, &sb, pglob)) 572 return(0); 573 574 if ((pglob->gl_flags & GLOB_LIMIT) && 575 limits[GLOB_INDEX_STAT]++ >= GLOB_LIMIT_STAT) { 576 errno = 0; 577 *pathend++ = SEP; 578 *pathend = EOS; 579 return GLOB_NOSPACE; 580 } 581 582 if (((pglob->gl_flags & GLOB_MARK) && 583 pathend[-1] != SEP) && (S_ISDIR(sb.st_mode) 584 || (S_ISLNK(sb.st_mode) && 585 (g_stat(pathbuf, &sb, pglob) == 0) && 586 S_ISDIR(sb.st_mode)))) { 587 if (pathend + 1 > pathend_last) 588 return (GLOB_ABORTED); 589 *pathend++ = SEP; 590 *pathend = EOS; 591 } 592 ++pglob->gl_matchc; 593 return(globextend(pathbuf, pglob, limit)); 594 } 595 596 /* Find end of next segment, copy tentatively to pathend. */ 597 q = pathend; 598 p = pattern; 599 while (*p != EOS && *p != SEP) { 600 if (ismeta(*p)) 601 anymeta = 1; 602 if (q + 1 > pathend_last) 603 return (GLOB_ABORTED); 604 *q++ = *p++; 605 } 606 607 if (!anymeta) { /* No expansion, do next segment. */ 608 pathend = q; 609 pattern = p; 610 while (*pattern == SEP) { 611 if (pathend + 1 > pathend_last) 612 return (GLOB_ABORTED); 613 *pathend++ = *pattern++; 614 } 615 } else /* Need expansion, recurse. */ 616 return(glob3(pathbuf, pathend, pathend_last, pattern, p, 617 pglob, limit)); 618 } 619 /* NOTREACHED */ 620 } 621 622 static int 623 glob3(Char *pathbuf, Char *pathend, Char *pathend_last, const Char *pattern, 624 const Char *restpattern, glob_t *pglob, size_t *limit) 625 { 626 struct dirent *dp; 627 DIR *dirp; 628 int err; 629 char buf[MAXPATHLEN]; 630 struct dirent *(*readdirfunc)(DIR *); 631 632 if (pathend > pathend_last) 633 return (GLOB_ABORTED); 634 *pathend = EOS; 635 errno = 0; 636 637 if ((dirp = g_opendir(pathbuf, pglob)) == NULL) { 638 /* TODO: don't call for ENOENT or ENOTDIR? */ 639 if (pglob->gl_errfunc) { 640 if (g_Ctoc(pathbuf, buf, sizeof(buf))) 641 return (GLOB_ABORTED); 642 if (pglob->gl_errfunc(buf, errno) || 643 pglob->gl_flags & GLOB_ERR) 644 return (GLOB_ABORTED); 645 } 646 return(0); 647 } 648 649 err = 0; 650 651 /* pglob->gl_readdir takes a void *, fix this manually. */ 652 if (pglob->gl_flags & GLOB_ALTDIRFUNC) 653 readdirfunc = (struct dirent *(*)(DIR *))pglob->gl_readdir; 654 else 655 readdirfunc = readdir; 656 657 /* Search directory for matching names. */ 658 while ((dp = (*readdirfunc)(dirp)) != NULL) { 659 char *sc; 660 Char *dc; 661 wchar_t wc; 662 size_t clen; 663 mbstate_t mbs; 664 665 if ((pglob->gl_flags & GLOB_LIMIT) && 666 limits[GLOB_INDEX_READDIR]++ >= GLOB_LIMIT_READDIR) { 667 errno = 0; 668 *pathend++ = SEP; 669 *pathend = EOS; 670 return GLOB_NOSPACE; 671 } 672 673 /* Initial DOT must be matched literally. */ 674 if (dp->d_name[0] == DOT && *pattern != DOT && 675 !(pglob->gl_flags & GLOB_PERIOD)) 676 continue; 677 memset(&mbs, 0, sizeof(mbs)); 678 dc = pathend; 679 sc = dp->d_name; 680 while (dc < pathend_last) { 681 clen = mbrtowc(&wc, sc, MB_LEN_MAX, &mbs); 682 if (clen == (size_t)-1 || clen == (size_t)-2) { 683 wc = *sc; 684 clen = 1; 685 memset(&mbs, 0, sizeof(mbs)); 686 } 687 if ((*dc++ = wc) == EOS) 688 break; 689 sc += clen; 690 } 691 if (!match(pathend, pattern, restpattern)) { 692 *pathend = EOS; 693 continue; 694 } 695 err = glob2(pathbuf, --dc, pathend_last, restpattern, 696 pglob, limit); 697 if (err) 698 break; 699 } 700 701 if (pglob->gl_flags & GLOB_ALTDIRFUNC) 702 (*pglob->gl_closedir)(dirp); 703 else 704 closedir(dirp); 705 return(err); 706 } 707 708 709 /* 710 * Extend the gl_pathv member of a glob_t structure to accomodate a new item, 711 * add the new item, and update gl_pathc. 712 * 713 * This assumes the BSD realloc, which only copies the block when its size 714 * crosses a power-of-two boundary; for v7 realloc, this would cause quadratic 715 * behavior. 716 * 717 * Return 0 if new item added, error code if memory couldn't be allocated. 718 * 719 * Invariant of the glob_t structure: 720 * Either gl_pathc is zero and gl_pathv is NULL; or gl_pathc > 0 and 721 * gl_pathv points to (gl_offs + gl_pathc + 1) items. 722 */ 723 static int 724 globextend(const Char *path, glob_t *pglob, size_t *limit) 725 { 726 char **pathv; 727 size_t i, newsize, len; 728 char *copy; 729 const Char *p; 730 731 if (*limit && pglob->gl_pathc > *limit) { 732 errno = 0; 733 return (GLOB_NOSPACE); 734 } 735 736 newsize = sizeof(*pathv) * (2 + pglob->gl_pathc + pglob->gl_offs); 737 pathv = pglob->gl_pathv ? 738 realloc((char *)pglob->gl_pathv, newsize) : 739 malloc(newsize); 740 if (pathv == NULL) { 741 if (pglob->gl_pathv) { 742 free(pglob->gl_pathv); 743 pglob->gl_pathv = NULL; 744 } 745 return(GLOB_NOSPACE); 746 } 747 748 if (pglob->gl_pathv == NULL && pglob->gl_offs > 0) { 749 /* first time around -- clear initial gl_offs items */ 750 pathv += pglob->gl_offs; 751 for (i = pglob->gl_offs + 1; --i > 0; ) 752 *--pathv = NULL; 753 } 754 pglob->gl_pathv = pathv; 755 756 for (p = path; *p++;) 757 continue; 758 len = MB_CUR_MAX * (size_t)(p - path); /* XXX overallocation */ 759 limits[GLOB_INDEX_MALLOC] += len; 760 if ((copy = malloc(len)) != NULL) { 761 if (g_Ctoc(path, copy, len)) { 762 free(copy); 763 return (GLOB_NOSPACE); 764 } 765 pathv[pglob->gl_offs + pglob->gl_pathc++] = copy; 766 } 767 pathv[pglob->gl_offs + pglob->gl_pathc] = NULL; 768 if ((pglob->gl_flags & GLOB_LIMIT) && 769 (newsize + limits[GLOB_INDEX_MALLOC]) >= GLOB_LIMIT_MALLOC) { 770 errno = 0; 771 return GLOB_NOSPACE; 772 } 773 774 return(copy == NULL ? GLOB_NOSPACE : 0); 775 } 776 777 /* 778 * pattern matching function for filenames. Each occurrence of the * 779 * pattern causes a recursion level. 780 */ 781 static int 782 match(const Char * name, const Char *pat, const Char *patend) 783 { 784 int ok, negate_range; 785 Char c, k; 786 787 while (pat < patend) { 788 c = *pat++; 789 switch (c & M_MASK) { 790 case M_ALL: 791 if (pat == patend) 792 return(1); 793 do 794 if (match(name, pat, patend)) 795 return(1); 796 while (*name++ != EOS); 797 return(0); 798 case M_ONE: 799 if (*name++ == EOS) 800 return(0); 801 break; 802 case M_SET: 803 ok = 0; 804 if ((k = *name++) == EOS) 805 return(0); 806 if ((negate_range = ((*pat & M_MASK) == M_NOT)) != EOS) 807 ++pat; 808 while (((c = *pat++) & M_MASK) != M_END) 809 if ((*pat & M_MASK) == M_RNG) { 810 if (__collate_load_error ? 811 CHAR(c) <= CHAR(k) && CHAR(k) <= CHAR(pat[1]) : 812 __collate_range_cmp(CHAR(c), CHAR(k)) <= 0 813 && __collate_range_cmp(CHAR(k), CHAR(pat[1])) <= 0 814 ) 815 ok = 1; 816 pat += 2; 817 } else if (c == k) 818 ok = 1; 819 if (ok == negate_range) 820 return(0); 821 break; 822 default: 823 if (*name++ != c) 824 return(0); 825 break; 826 } 827 } 828 return(*name == EOS); 829 } 830 831 /* Free allocated data belonging to a glob_t structure. */ 832 void 833 globfree(glob_t *pglob) 834 { 835 size_t i; 836 char **pp; 837 838 if (pglob->gl_pathv != NULL) { 839 pp = pglob->gl_pathv + pglob->gl_offs; 840 for (i = pglob->gl_pathc; i--; ++pp) 841 if (*pp) 842 free(*pp); 843 free(pglob->gl_pathv); 844 pglob->gl_pathv = NULL; 845 } 846 } 847 848 static DIR * 849 g_opendir(const Char *str, glob_t *pglob) 850 { 851 char buf[MAXPATHLEN]; 852 853 if (*str == '\0') 854 strcpy(buf, "."); 855 else if (g_Ctoc(str, buf, sizeof(buf))) 856 return (NULL); 857 858 if (pglob->gl_flags & GLOB_ALTDIRFUNC) 859 return((*pglob->gl_opendir)(buf)); 860 861 return(opendir(buf)); 862 } 863 864 static int 865 g_lstat(const Char *fn, struct stat *sb, glob_t *pglob) 866 { 867 char buf[MAXPATHLEN]; 868 869 if (g_Ctoc(fn, buf, sizeof(buf))) { 870 errno = ENAMETOOLONG; 871 return (-1); 872 } 873 if (pglob->gl_flags & GLOB_ALTDIRFUNC) 874 return((*pglob->gl_lstat)(buf, sb)); 875 return(lstat(buf, sb)); 876 } 877 878 static int 879 g_stat(const Char *fn, struct stat *sb, glob_t *pglob) 880 { 881 char buf[MAXPATHLEN]; 882 883 if (g_Ctoc(fn, buf, sizeof(buf))) { 884 errno = ENAMETOOLONG; 885 return (-1); 886 } 887 if (pglob->gl_flags & GLOB_ALTDIRFUNC) 888 return((*pglob->gl_stat)(buf, sb)); 889 return(stat(buf, sb)); 890 } 891 892 static const Char * 893 g_strchr(const Char *str, wchar_t ch) 894 { 895 for (;; ++str) { 896 if (*str == ch) 897 return(str); 898 if (*str == '\0') 899 return(NULL); 900 } 901 } 902 903 static int 904 g_Ctoc(const Char *str, char *buf, size_t len) 905 { 906 mbstate_t mbs; 907 size_t clen; 908 909 memset(&mbs, 0, sizeof(mbs)); 910 while (len >= MB_CUR_MAX) { 911 clen = wcrtomb(buf, *str, &mbs); 912 if (clen == (size_t)-1) 913 return (1); 914 if (*str == L'\0') 915 return (0); 916 str++; 917 buf += clen; 918 len -= clen; 919 } 920 return (1); 921 } 922 923 #ifdef DEBUG 924 static void 925 qprintf(const char *str, const Char *s) 926 { 927 const Char *p; 928 929 printf("%s:\n", str); 930 for (p = s; *p; p++) 931 printf("%c", CHAR(*p)); 932 printf("\n"); 933 for (p = s; *p; p++) 934 printf("%c", *p & M_PROTECT ? '"' : ' '); 935 printf("\n"); 936 for (p = s; *p; p++) 937 printf("%c", ismeta(*p) ? '_' : ' '); 938 printf("\n"); 939 } 940 #endif 941