| 1 | /* |
| 2 | * Natanael Arndt, 2011: removed collate.h dependencies |
| 3 | * (my changes are trivial) |
| 4 | * |
| 5 | * Copyright (c) 1989, 1993 |
| 6 | * The Regents of the University of California. All rights reserved. |
| 7 | * |
| 8 | * This code is derived from software contributed to Berkeley by |
| 9 | * Guido van Rossum. |
| 10 | * |
| 11 | * Redistribution and use in source and binary forms, with or without |
| 12 | * modification, are permitted provided that the following conditions |
| 13 | * are met: |
| 14 | * 1. Redistributions of source code must retain the above copyright |
| 15 | * notice, this list of conditions and the following disclaimer. |
| 16 | * 2. Redistributions in binary form must reproduce the above copyright |
| 17 | * notice, this list of conditions and the following disclaimer in the |
| 18 | * documentation and/or other materials provided with the distribution. |
| 19 | * 4. Neither the name of the University nor the names of its contributors |
| 20 | * may be used to endorse or promote products derived from this software |
| 21 | * without specific prior written permission. |
| 22 | * |
| 23 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
| 24 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 25 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 26 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
| 27 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
| 28 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
| 29 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
| 30 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
| 31 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
| 32 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 33 | * SUCH DAMAGE. |
| 34 | */ |
| 35 | |
| 36 | #if defined(LIBC_SCCS) && !defined(lint) |
| 37 | static char sccsid[] = "@(#)glob.c 8.3 (Berkeley) 10/13/93"; |
| 38 | #endif /* LIBC_SCCS and not lint */ |
| 39 | #include <sys/cdefs.h> |
| 40 | __FBSDID("$FreeBSD$"); |
| 41 | |
| 42 | /* |
| 43 | * glob(3) -- a superset of the one defined in POSIX 1003.2. |
| 44 | * |
| 45 | * The [!...] convention to negate a range is supported (SysV, Posix, ksh). |
| 46 | * |
| 47 | * Optional extra services, controlled by flags not defined by POSIX: |
| 48 | * |
| 49 | * GLOB_QUOTE: |
| 50 | * Escaping convention: \ inhibits any special meaning the following |
| 51 | * character might have (except \ at end of string is retained). |
| 52 | * GLOB_MAGCHAR: |
| 53 | * Set in gl_flags if pattern contained a globbing character. |
| 54 | * GLOB_NOMAGIC: |
| 55 | * Same as GLOB_NOCHECK, but it will only append pattern if it did |
| 56 | * not contain any magic characters. [Used in csh style globbing] |
| 57 | * GLOB_ALTDIRFUNC: |
| 58 | * Use alternately specified directory access functions. |
| 59 | * GLOB_TILDE: |
| 60 | * expand ~user/foo to the /home/dir/of/user/foo |
| 61 | * GLOB_BRACE: |
| 62 | * expand {1,2}{a,b} to 1a 1b 2a 2b |
| 63 | * gl_matchc: |
| 64 | * Number of matches in the current invocation of glob. |
| 65 | */ |
| 66 | |
| 67 | /* |
| 68 | * Some notes on multibyte character support: |
| 69 | * 1. Patterns with illegal byte sequences match nothing - even if |
| 70 | * GLOB_NOCHECK is specified. |
| 71 | * 2. Illegal byte sequences in filenames are handled by treating them as |
| 72 | * single-byte characters with a value of the first byte of the sequence |
| 73 | * cast to wchar_t. |
| 74 | * 3. State-dependent encodings are not currently supported. |
| 75 | */ |
| 76 | |
| 77 | #include <sys/param.h> |
| 78 | #include <sys/stat.h> |
| 79 | |
| 80 | #include <ctype.h> |
| 81 | #include <dirent.h> |
| 82 | #include <errno.h> |
| 83 | #include <glob.h> |
| 84 | #include <limits.h> |
| 85 | #include <pwd.h> |
| 86 | #include <stdint.h> |
| 87 | #include <stdio.h> |
| 88 | #include <stdlib.h> |
| 89 | #include <string.h> |
| 90 | #include <unistd.h> |
| 91 | #include <wchar.h> |
| 92 | |
| 93 | #define DOLLAR '$' |
| 94 | #define DOT '.' |
| 95 | #define EOS '\0' |
| 96 | #define LBRACKET '[' |
| 97 | #define NOT '!' |
| 98 | #define QUESTION '?' |
| 99 | #define QUOTE '\\' |
| 100 | #define RANGE '-' |
| 101 | #define RBRACKET ']' |
| 102 | #define SEP '/' |
| 103 | #define STAR '*' |
| 104 | #define TILDE '~' |
| 105 | #define UNDERSCORE '_' |
| 106 | #define LBRACE '{' |
| 107 | #define RBRACE '}' |
| 108 | #define SLASH '/' |
| 109 | #define COMMA ',' |
| 110 | |
| 111 | #ifndef DEBUG |
| 112 | |
| 113 | #define M_QUOTE 0x8000000000ULL |
| 114 | #define M_PROTECT 0x4000000000ULL |
| 115 | #define M_MASK 0xffffffffffULL |
| 116 | #define M_CHAR 0x00ffffffffULL |
| 117 | |
| 118 | typedef uint_fast64_t Char; |
| 119 | |
| 120 | #else |
| 121 | |
| 122 | #define M_QUOTE 0x80 |
| 123 | #define M_PROTECT 0x40 |
| 124 | #define M_MASK 0xff |
| 125 | #define M_CHAR 0x7f |
| 126 | |
| 127 | typedef char Char; |
| 128 | |
| 129 | #endif |
| 130 | |
| 131 | |
| 132 | #define CHAR(c) ((Char)((c)&M_CHAR)) |
| 133 | #define META(c) ((Char)((c)|M_QUOTE)) |
| 134 | #define M_ALL META('*') |
| 135 | #define M_END META(']') |
| 136 | #define M_NOT META('!') |
| 137 | #define M_ONE META('?') |
| 138 | #define M_RNG META('-') |
| 139 | #define M_SET META('[') |
| 140 | #define ismeta(c) (((c)&M_QUOTE) != 0) |
| 141 | |
| 142 | |
| 143 | static int compare(const void *, const void *); |
| 144 | static int g_Ctoc(const Char *, char *, size_t); |
| 145 | static int g_lstat(Char *, struct stat *, glob_t *); |
| 146 | static DIR *g_opendir(Char *, glob_t *); |
| 147 | static const Char *g_strchr(const Char *, wchar_t); |
| 148 | #ifdef notdef |
| 149 | static Char *g_strcat(Char *, const Char *); |
| 150 | #endif |
| 151 | static int g_stat(Char *, struct stat *, glob_t *); |
| 152 | static int glob0(const Char *, glob_t *, size_t *); |
| 153 | static int glob1(Char *, glob_t *, size_t *); |
| 154 | static int glob2(Char *, Char *, Char *, Char *, glob_t *, size_t *); |
| 155 | static int glob3(Char *, Char *, Char *, Char *, Char *, glob_t *, size_t *); |
| 156 | static int globextend(const Char *, glob_t *, size_t *); |
| 157 | static const Char * |
| 158 | globtilde(const Char *, Char *, size_t, glob_t *); |
| 159 | static int globexp1(const Char *, glob_t *, size_t *); |
| 160 | static int globexp2(const Char *, const Char *, glob_t *, int *, size_t *); |
| 161 | static int match(Char *, Char *, Char *); |
| 162 | #ifdef DEBUG |
| 163 | static void qprintf(const char *, Char *); |
| 164 | #endif |
| 165 | |
| 166 | int |
| 167 | glob(const char *pattern, int flags, int (*errfunc)(const char *, int), glob_t *pglob) |
| 168 | { |
| 169 | const char *patnext; |
| 170 | size_t limit; |
| 171 | Char *bufnext, *bufend, patbuf[MAXPATHLEN], prot; |
| 172 | mbstate_t mbs; |
| 173 | wchar_t wc; |
| 174 | size_t clen; |
| 175 | |
| 176 | patnext = pattern; |
| 177 | if (!(flags & GLOB_APPEND)) { |
| 178 | pglob->gl_pathc = 0; |
| 179 | pglob->gl_pathv = NULL; |
| 180 | if (!(flags & GLOB_DOOFFS)) |
| 181 | pglob->gl_offs = 0; |
| 182 | } |
| 183 | if (flags & GLOB_LIMIT) { |
| 184 | limit = pglob->gl_matchc; |
| 185 | if (limit == 0) |
| 186 | limit = ARG_MAX; |
| 187 | } else |
| 188 | limit = 0; |
| 189 | pglob->gl_flags = flags & ~GLOB_MAGCHAR; |
| 190 | pglob->gl_errfunc = errfunc; |
| 191 | pglob->gl_matchc = 0; |
| 192 | |
| 193 | bufnext = patbuf; |
| 194 | bufend = bufnext + MAXPATHLEN - 1; |
| 195 | if (flags & GLOB_NOESCAPE) { |
| 196 | memset(&mbs, 0, sizeof(mbs)); |
| 197 | while (bufend - bufnext >= MB_CUR_MAX) { |
| 198 | clen = mbrtowc(&wc, patnext, MB_LEN_MAX, &mbs); |
| 199 | if (clen == (size_t)-1 || clen == (size_t)-2) |
| 200 | return (GLOB_NOMATCH); |
| 201 | else if (clen == 0) |
| 202 | break; |
| 203 | *bufnext++ = wc; |
| 204 | patnext += clen; |
| 205 | } |
| 206 | } else { |
| 207 | /* Protect the quoted characters. */ |
| 208 | memset(&mbs, 0, sizeof(mbs)); |
| 209 | while (bufend - bufnext >= MB_CUR_MAX) { |
| 210 | if (*patnext == QUOTE) { |
| 211 | if (*++patnext == EOS) { |
| 212 | *bufnext++ = QUOTE | M_PROTECT; |
| 213 | continue; |
| 214 | } |
| 215 | prot = M_PROTECT; |
| 216 | } else |
| 217 | prot = 0; |
| 218 | clen = mbrtowc(&wc, patnext, MB_LEN_MAX, &mbs); |
| 219 | if (clen == (size_t)-1 || clen == (size_t)-2) |
| 220 | return (GLOB_NOMATCH); |
| 221 | else if (clen == 0) |
| 222 | break; |
| 223 | *bufnext++ = wc | prot; |
| 224 | patnext += clen; |
| 225 | } |
| 226 | } |
| 227 | *bufnext = EOS; |
| 228 | |
| 229 | if (flags & GLOB_BRACE) |
| 230 | return globexp1(patbuf, pglob, &limit); |
| 231 | else |
| 232 | return glob0(patbuf, pglob, &limit); |
| 233 | } |
| 234 | |
| 235 | /* |
| 236 | * Expand recursively a glob {} pattern. When there is no more expansion |
| 237 | * invoke the standard globbing routine to glob the rest of the magic |
| 238 | * characters |
| 239 | */ |
| 240 | static int |
| 241 | globexp1(const Char *pattern, glob_t *pglob, size_t *limit) |
| 242 | { |
| 243 | const Char* ptr = pattern; |
| 244 | int rv; |
| 245 | |
| 246 | /* Protect a single {}, for find(1), like csh */ |
| 247 | if (pattern[0] == LBRACE && pattern[1] == RBRACE && pattern[2] == EOS) |
| 248 | return glob0(pattern, pglob, limit); |
| 249 | |
| 250 | while ((ptr = g_strchr(ptr, LBRACE)) != NULL) |
| 251 | if (!globexp2(ptr, pattern, pglob, &rv, limit)) |
| 252 | return rv; |
| 253 | |
| 254 | return glob0(pattern, pglob, limit); |
| 255 | } |
| 256 | |
| 257 | |
| 258 | /* |
| 259 | * Recursive brace globbing helper. Tries to expand a single brace. |
| 260 | * If it succeeds then it invokes globexp1 with the new pattern. |
| 261 | * If it fails then it tries to glob the rest of the pattern and returns. |
| 262 | */ |
| 263 | static int |
| 264 | globexp2(const Char *ptr, const Char *pattern, glob_t *pglob, int *rv, size_t *limit) |
| 265 | { |
| 266 | int i; |
| 267 | Char *lm, *ls; |
| 268 | const Char *pe, *pm, *pm1, *pl; |
| 269 | Char patbuf[MAXPATHLEN]; |
| 270 | |
| 271 | /* copy part up to the brace */ |
| 272 | for (lm = patbuf, pm = pattern; pm != ptr; *lm++ = *pm++) |
| 273 | continue; |
| 274 | *lm = EOS; |
| 275 | ls = lm; |
| 276 | |
| 277 | /* Find the balanced brace */ |
| 278 | for (i = 0, pe = ++ptr; *pe; pe++) |
| 279 | if (*pe == LBRACKET) { |
| 280 | /* Ignore everything between [] */ |
| 281 | for (pm = pe++; *pe != RBRACKET && *pe != EOS; pe++) |
| 282 | continue; |
| 283 | if (*pe == EOS) { |
| 284 | /* |
| 285 | * We could not find a matching RBRACKET. |
| 286 | * Ignore and just look for RBRACE |
| 287 | */ |
| 288 | pe = pm; |
| 289 | } |
| 290 | } |
| 291 | else if (*pe == LBRACE) |
| 292 | i++; |
| 293 | else if (*pe == RBRACE) { |
| 294 | if (i == 0) |
| 295 | break; |
| 296 | i--; |
| 297 | } |
| 298 | |
| 299 | /* Non matching braces; just glob the pattern */ |
| 300 | if (i != 0 || *pe == EOS) { |
| 301 | *rv = glob0(patbuf, pglob, limit); |
| 302 | return 0; |
| 303 | } |
| 304 | |
| 305 | for (i = 0, pl = pm = ptr; pm <= pe; pm++) |
| 306 | switch (*pm) { |
| 307 | case LBRACKET: |
| 308 | /* Ignore everything between [] */ |
| 309 | for (pm1 = pm++; *pm != RBRACKET && *pm != EOS; pm++) |
| 310 | continue; |
| 311 | if (*pm == EOS) { |
| 312 | /* |
| 313 | * We could not find a matching RBRACKET. |
| 314 | * Ignore and just look for RBRACE |
| 315 | */ |
| 316 | pm = pm1; |
| 317 | } |
| 318 | break; |
| 319 | |
| 320 | case LBRACE: |
| 321 | i++; |
| 322 | break; |
| 323 | |
| 324 | case RBRACE: |
| 325 | if (i) { |
| 326 | i--; |
| 327 | break; |
| 328 | } |
| 329 | /* FALLTHROUGH */ |
| 330 | case COMMA: |
| 331 | if (i && *pm == COMMA) |
| 332 | break; |
| 333 | else { |
| 334 | /* Append the current string */ |
| 335 | for (lm = ls; (pl < pm); *lm++ = *pl++) |
| 336 | continue; |
| 337 | /* |
| 338 | * Append the rest of the pattern after the |
| 339 | * closing brace |
| 340 | */ |
| 341 | for (pl = pe + 1; (*lm++ = *pl++) != EOS;) |
| 342 | continue; |
| 343 | |
| 344 | /* Expand the current pattern */ |
| 345 | #ifdef DEBUG |
| 346 | qprintf("globexp2:", patbuf); |
| 347 | #endif |
| 348 | *rv = globexp1(patbuf, pglob, limit); |
| 349 | |
| 350 | /* move after the comma, to the next string */ |
| 351 | pl = pm + 1; |
| 352 | } |
| 353 | break; |
| 354 | |
| 355 | default: |
| 356 | break; |
| 357 | } |
| 358 | *rv = 0; |
| 359 | return 0; |
| 360 | } |
| 361 | |
| 362 | |
| 363 | |
| 364 | /* |
| 365 | * expand tilde from the passwd file. |
| 366 | */ |
| 367 | static const Char * |
| 368 | globtilde(const Char *pattern, Char *patbuf, size_t patbuf_len, glob_t *pglob) |
| 369 | { |
| 370 | struct passwd *pwd; |
| 371 | char *h; |
| 372 | const Char *p; |
| 373 | Char *b, *eb; |
| 374 | |
| 375 | if (*pattern != TILDE || !(pglob->gl_flags & GLOB_TILDE)) |
| 376 | return pattern; |
| 377 | |
| 378 | /* |
| 379 | * Copy up to the end of the string or / |
| 380 | */ |
| 381 | eb = &patbuf[patbuf_len - 1]; |
| 382 | for (p = pattern + 1, h = (char *) patbuf; |
| 383 | h < (char *)eb && *p && *p != SLASH; *h++ = *p++) |
| 384 | continue; |
| 385 | |
| 386 | *h = EOS; |
| 387 | |
| 388 | if (((char *) patbuf)[0] == EOS) { |
| 389 | /* |
| 390 | * handle a plain ~ or ~/ by expanding $HOME first (iff |
| 391 | * we're not running setuid or setgid) and then trying |
| 392 | * the password file |
| 393 | */ |
| 394 | if ( |
| 395 | #ifndef __ANDROID__ |
| 396 | issetugid() != 0 || |
| 397 | #endif |
| 398 | (h = getenv("HOME")) == NULL) { |
| 399 | if (((h = getlogin()) != NULL && |
| 400 | (pwd = getpwnam(h)) != NULL) || |
| 401 | (pwd = getpwuid(getuid())) != NULL) |
| 402 | h = pwd->pw_dir; |
| 403 | else |
| 404 | return pattern; |
| 405 | } |
| 406 | } |
| 407 | else { |
| 408 | /* |
| 409 | * Expand a ~user |
| 410 | */ |
| 411 | if ((pwd = getpwnam((char*) patbuf)) == NULL) |
| 412 | return pattern; |
| 413 | else |
| 414 | h = pwd->pw_dir; |
| 415 | } |
| 416 | |
| 417 | /* Copy the home directory */ |
| 418 | for (b = patbuf; b < eb && *h; *b++ = *h++) |
| 419 | continue; |
| 420 | |
| 421 | /* Append the rest of the pattern */ |
| 422 | while (b < eb && (*b++ = *p++) != EOS) |
| 423 | continue; |
| 424 | *b = EOS; |
| 425 | |
| 426 | return patbuf; |
| 427 | } |
| 428 | |
| 429 | |
| 430 | /* |
| 431 | * The main glob() routine: compiles the pattern (optionally processing |
| 432 | * quotes), calls glob1() to do the real pattern matching, and finally |
| 433 | * sorts the list (unless unsorted operation is requested). Returns 0 |
| 434 | * if things went well, nonzero if errors occurred. |
| 435 | */ |
| 436 | static int |
| 437 | glob0(const Char *pattern, glob_t *pglob, size_t *limit) |
| 438 | { |
| 439 | const Char *qpatnext; |
| 440 | int err; |
| 441 | size_t oldpathc; |
| 442 | Char *bufnext, c, patbuf[MAXPATHLEN]; |
| 443 | |
| 444 | qpatnext = globtilde(pattern, patbuf, MAXPATHLEN, pglob); |
| 445 | oldpathc = pglob->gl_pathc; |
| 446 | bufnext = patbuf; |
| 447 | |
| 448 | /* We don't need to check for buffer overflow any more. */ |
| 449 | while ((c = *qpatnext++) != EOS) { |
| 450 | switch (c) { |
| 451 | case LBRACKET: |
| 452 | c = *qpatnext; |
| 453 | if (c == NOT) |
| 454 | ++qpatnext; |
| 455 | if (*qpatnext == EOS || |
| 456 | g_strchr(qpatnext+1, RBRACKET) == NULL) { |
| 457 | *bufnext++ = LBRACKET; |
| 458 | if (c == NOT) |
| 459 | --qpatnext; |
| 460 | break; |
| 461 | } |
| 462 | *bufnext++ = M_SET; |
| 463 | if (c == NOT) |
| 464 | *bufnext++ = M_NOT; |
| 465 | c = *qpatnext++; |
| 466 | do { |
| 467 | *bufnext++ = CHAR(c); |
| 468 | if (*qpatnext == RANGE && |
| 469 | (c = qpatnext[1]) != RBRACKET) { |
| 470 | *bufnext++ = M_RNG; |
| 471 | *bufnext++ = CHAR(c); |
| 472 | qpatnext += 2; |
| 473 | } |
| 474 | } while ((c = *qpatnext++) != RBRACKET); |
| 475 | pglob->gl_flags |= GLOB_MAGCHAR; |
| 476 | *bufnext++ = M_END; |
| 477 | break; |
| 478 | case QUESTION: |
| 479 | pglob->gl_flags |= GLOB_MAGCHAR; |
| 480 | *bufnext++ = M_ONE; |
| 481 | break; |
| 482 | case STAR: |
| 483 | pglob->gl_flags |= GLOB_MAGCHAR; |
| 484 | /* collapse adjacent stars to one, |
| 485 | * to avoid exponential behavior |
| 486 | */ |
| 487 | if (bufnext == patbuf || bufnext[-1] != M_ALL) |
| 488 | *bufnext++ = M_ALL; |
| 489 | break; |
| 490 | default: |
| 491 | *bufnext++ = CHAR(c); |
| 492 | break; |
| 493 | } |
| 494 | } |
| 495 | *bufnext = EOS; |
| 496 | #ifdef DEBUG |
| 497 | qprintf("glob0:", patbuf); |
| 498 | #endif |
| 499 | |
| 500 | if ((err = glob1(patbuf, pglob, limit)) != 0) |
| 501 | return(err); |
| 502 | |
| 503 | /* |
| 504 | * If there was no match we are going to append the pattern |
| 505 | * if GLOB_NOCHECK was specified or if GLOB_NOMAGIC was specified |
| 506 | * and the pattern did not contain any magic characters |
| 507 | * GLOB_NOMAGIC is there just for compatibility with csh. |
| 508 | */ |
| 509 | if (pglob->gl_pathc == oldpathc) { |
| 510 | if (((pglob->gl_flags & GLOB_NOCHECK) || |
| 511 | ((pglob->gl_flags & GLOB_NOMAGIC) && |
| 512 | !(pglob->gl_flags & GLOB_MAGCHAR)))) |
| 513 | return(globextend(pattern, pglob, limit)); |
| 514 | else |
| 515 | return(GLOB_NOMATCH); |
| 516 | } |
| 517 | if (!(pglob->gl_flags & GLOB_NOSORT)) |
| 518 | qsort(pglob->gl_pathv + pglob->gl_offs + oldpathc, |
| 519 | pglob->gl_pathc - oldpathc, sizeof(char *), compare); |
| 520 | return(0); |
| 521 | } |
| 522 | |
| 523 | static int |
| 524 | compare(const void *p, const void *q) |
| 525 | { |
| 526 | return(strcmp(*(char **)p, *(char **)q)); |
| 527 | } |
| 528 | |
| 529 | static int |
| 530 | glob1(Char *pattern, glob_t *pglob, size_t *limit) |
| 531 | { |
| 532 | Char pathbuf[MAXPATHLEN]; |
| 533 | |
| 534 | /* A null pathname is invalid -- POSIX 1003.1 sect. 2.4. */ |
| 535 | if (*pattern == EOS) |
| 536 | return(0); |
| 537 | return(glob2(pathbuf, pathbuf, pathbuf + MAXPATHLEN - 1, |
| 538 | pattern, pglob, limit)); |
| 539 | } |
| 540 | |
| 541 | /* |
| 542 | * The functions glob2 and glob3 are mutually recursive; there is one level |
| 543 | * of recursion for each segment in the pattern that contains one or more |
| 544 | * meta characters. |
| 545 | */ |
| 546 | static int |
| 547 | glob2(Char *pathbuf, Char *pathend, Char *pathend_last, Char *pattern, |
| 548 | glob_t *pglob, size_t *limit) |
| 549 | { |
| 550 | struct stat sb; |
| 551 | Char *p, *q; |
| 552 | int anymeta; |
| 553 | |
| 554 | /* |
| 555 | * Loop over pattern segments until end of pattern or until |
| 556 | * segment with meta character found. |
| 557 | */ |
| 558 | for (anymeta = 0;;) { |
| 559 | if (*pattern == EOS) { /* End of pattern? */ |
| 560 | *pathend = EOS; |
| 561 | if (g_lstat(pathbuf, &sb, pglob)) |
| 562 | return(0); |
| 563 | |
| 564 | if (((pglob->gl_flags & GLOB_MARK) && |
| 565 | pathend[-1] != SEP) && (S_ISDIR(sb.st_mode) |
| 566 | || (S_ISLNK(sb.st_mode) && |
| 567 | (g_stat(pathbuf, &sb, pglob) == 0) && |
| 568 | S_ISDIR(sb.st_mode)))) { |
| 569 | if (pathend + 1 > pathend_last) |
| 570 | return (GLOB_ABORTED); |
| 571 | *pathend++ = SEP; |
| 572 | *pathend = EOS; |
| 573 | } |
| 574 | ++pglob->gl_matchc; |
| 575 | return(globextend(pathbuf, pglob, limit)); |
| 576 | } |
| 577 | |
| 578 | /* Find end of next segment, copy tentatively to pathend. */ |
| 579 | q = pathend; |
| 580 | p = pattern; |
| 581 | while (*p != EOS && *p != SEP) { |
| 582 | if (ismeta(*p)) |
| 583 | anymeta = 1; |
| 584 | if (q + 1 > pathend_last) |
| 585 | return (GLOB_ABORTED); |
| 586 | *q++ = *p++; |
| 587 | } |
| 588 | |
| 589 | if (!anymeta) { /* No expansion, do next segment. */ |
| 590 | pathend = q; |
| 591 | pattern = p; |
| 592 | while (*pattern == SEP) { |
| 593 | if (pathend + 1 > pathend_last) |
| 594 | return (GLOB_ABORTED); |
| 595 | *pathend++ = *pattern++; |
| 596 | } |
| 597 | } else /* Need expansion, recurse. */ |
| 598 | return(glob3(pathbuf, pathend, pathend_last, pattern, p, |
| 599 | pglob, limit)); |
| 600 | } |
| 601 | /* NOTREACHED */ |
| 602 | } |
| 603 | |
| 604 | static int |
| 605 | glob3(Char *pathbuf, Char *pathend, Char *pathend_last, |
| 606 | Char *pattern, Char *restpattern, |
| 607 | glob_t *pglob, size_t *limit) |
| 608 | { |
| 609 | struct dirent *dp; |
| 610 | DIR *dirp; |
| 611 | int err; |
| 612 | char buf[MAXPATHLEN]; |
| 613 | |
| 614 | /* |
| 615 | * The readdirfunc declaration can't be prototyped, because it is |
| 616 | * assigned, below, to two functions which are prototyped in glob.h |
| 617 | * and dirent.h as taking pointers to differently typed opaque |
| 618 | * structures. |
| 619 | */ |
| 620 | struct dirent *(*readdirfunc)(); |
| 621 | |
| 622 | if (pathend > pathend_last) |
| 623 | return (GLOB_ABORTED); |
| 624 | *pathend = EOS; |
| 625 | errno = 0; |
| 626 | |
| 627 | if ((dirp = g_opendir(pathbuf, pglob)) == NULL) { |
| 628 | /* TODO: don't call for ENOENT or ENOTDIR? */ |
| 629 | if (pglob->gl_errfunc) { |
| 630 | if (g_Ctoc(pathbuf, buf, sizeof(buf))) |
| 631 | return (GLOB_ABORTED); |
| 632 | if (pglob->gl_errfunc(buf, errno) || |
| 633 | pglob->gl_flags & GLOB_ERR) |
| 634 | return (GLOB_ABORTED); |
| 635 | } |
| 636 | return(0); |
| 637 | } |
| 638 | |
| 639 | err = 0; |
| 640 | |
| 641 | /* Search directory for matching names. */ |
| 642 | if (pglob->gl_flags & GLOB_ALTDIRFUNC) |
| 643 | readdirfunc = pglob->gl_readdir; |
| 644 | else |
| 645 | readdirfunc = readdir; |
| 646 | while ((dp = (*readdirfunc)(dirp))) { |
| 647 | char *sc; |
| 648 | Char *dc; |
| 649 | wchar_t wc; |
| 650 | size_t clen; |
| 651 | mbstate_t mbs; |
| 652 | |
| 653 | /* Initial DOT must be matched literally. */ |
| 654 | if (dp->d_name[0] == DOT && *pattern != DOT) |
| 655 | continue; |
| 656 | memset(&mbs, 0, sizeof(mbs)); |
| 657 | dc = pathend; |
| 658 | sc = dp->d_name; |
| 659 | while (dc < pathend_last) { |
| 660 | clen = mbrtowc(&wc, sc, MB_LEN_MAX, &mbs); |
| 661 | if (clen == (size_t)-1 || clen == (size_t)-2) { |
| 662 | wc = *sc; |
| 663 | clen = 1; |
| 664 | memset(&mbs, 0, sizeof(mbs)); |
| 665 | } |
| 666 | if ((*dc++ = wc) == EOS) |
| 667 | break; |
| 668 | sc += clen; |
| 669 | } |
| 670 | if (!match(pathend, pattern, restpattern)) { |
| 671 | *pathend = EOS; |
| 672 | continue; |
| 673 | } |
| 674 | err = glob2(pathbuf, --dc, pathend_last, restpattern, |
| 675 | pglob, limit); |
| 676 | if (err) |
| 677 | break; |
| 678 | } |
| 679 | |
| 680 | if (pglob->gl_flags & GLOB_ALTDIRFUNC) |
| 681 | (*pglob->gl_closedir)(dirp); |
| 682 | else |
| 683 | closedir(dirp); |
| 684 | return(err); |
| 685 | } |
| 686 | |
| 687 | |
| 688 | /* |
| 689 | * Extend the gl_pathv member of a glob_t structure to accomodate a new item, |
| 690 | * add the new item, and update gl_pathc. |
| 691 | * |
| 692 | * This assumes the BSD realloc, which only copies the block when its size |
| 693 | * crosses a power-of-two boundary; for v7 realloc, this would cause quadratic |
| 694 | * behavior. |
| 695 | * |
| 696 | * Return 0 if new item added, error code if memory couldn't be allocated. |
| 697 | * |
| 698 | * Invariant of the glob_t structure: |
| 699 | * Either gl_pathc is zero and gl_pathv is NULL; or gl_pathc > 0 and |
| 700 | * gl_pathv points to (gl_offs + gl_pathc + 1) items. |
| 701 | */ |
| 702 | static int |
| 703 | globextend(const Char *path, glob_t *pglob, size_t *limit) |
| 704 | { |
| 705 | char **pathv; |
| 706 | size_t i, newsize, len; |
| 707 | char *copy; |
| 708 | const Char *p; |
| 709 | |
| 710 | if (*limit && pglob->gl_pathc > *limit) { |
| 711 | errno = 0; |
| 712 | return (GLOB_NOSPACE); |
| 713 | } |
| 714 | |
| 715 | newsize = sizeof(*pathv) * (2 + pglob->gl_pathc + pglob->gl_offs); |
| 716 | pathv = pglob->gl_pathv ? |
| 717 | realloc((char *)pglob->gl_pathv, newsize) : |
| 718 | malloc(newsize); |
| 719 | if (pathv == NULL) { |
| 720 | if (pglob->gl_pathv) { |
| 721 | free(pglob->gl_pathv); |
| 722 | pglob->gl_pathv = NULL; |
| 723 | } |
| 724 | return(GLOB_NOSPACE); |
| 725 | } |
| 726 | |
| 727 | if (pglob->gl_pathv == NULL && pglob->gl_offs > 0) { |
| 728 | /* first time around -- clear initial gl_offs items */ |
| 729 | pathv += pglob->gl_offs; |
| 730 | for (i = pglob->gl_offs + 1; --i > 0; ) |
| 731 | *--pathv = NULL; |
| 732 | } |
| 733 | pglob->gl_pathv = pathv; |
| 734 | |
| 735 | for (p = path; *p++;) |
| 736 | continue; |
| 737 | len = MB_CUR_MAX * (size_t)(p - path); /* XXX overallocation */ |
| 738 | if ((copy = malloc(len)) != NULL) { |
| 739 | if (g_Ctoc(path, copy, len)) { |
| 740 | free(copy); |
| 741 | return (GLOB_NOSPACE); |
| 742 | } |
| 743 | pathv[pglob->gl_offs + pglob->gl_pathc++] = copy; |
| 744 | } |
| 745 | pathv[pglob->gl_offs + pglob->gl_pathc] = NULL; |
| 746 | return(copy == NULL ? GLOB_NOSPACE : 0); |
| 747 | } |
| 748 | |
| 749 | /* |
| 750 | * pattern matching function for filenames. Each occurrence of the * |
| 751 | * pattern causes a recursion level. |
| 752 | */ |
| 753 | static int |
| 754 | match(Char *name, Char *pat, Char *patend) |
| 755 | { |
| 756 | int ok, negate_range; |
| 757 | Char c, k; |
| 758 | |
| 759 | while (pat < patend) { |
| 760 | c = *pat++; |
| 761 | switch (c & M_MASK) { |
| 762 | case M_ALL: |
| 763 | if (pat == patend) |
| 764 | return(1); |
| 765 | do |
| 766 | if (match(name, pat, patend)) |
| 767 | return(1); |
| 768 | while (*name++ != EOS); |
| 769 | return(0); |
| 770 | case M_ONE: |
| 771 | if (*name++ == EOS) |
| 772 | return(0); |
| 773 | break; |
| 774 | case M_SET: |
| 775 | ok = 0; |
| 776 | if ((k = *name++) == EOS) |
| 777 | return(0); |
| 778 | if ((negate_range = ((*pat & M_MASK) == M_NOT)) != EOS) |
| 779 | ++pat; |
| 780 | while (((c = *pat++) & M_MASK) != M_END) |
| 781 | if ((*pat & M_MASK) == M_RNG) { |
| 782 | if (CHAR(c) <= CHAR(k) && CHAR(k) <= CHAR(pat[1])) ok = 1; |
| 783 | pat += 2; |
| 784 | } else if (c == k) |
| 785 | ok = 1; |
| 786 | if (ok == negate_range) |
| 787 | return(0); |
| 788 | break; |
| 789 | default: |
| 790 | if (*name++ != c) |
| 791 | return(0); |
| 792 | break; |
| 793 | } |
| 794 | } |
| 795 | return(*name == EOS); |
| 796 | } |
| 797 | |
| 798 | /* Free allocated data belonging to a glob_t structure. */ |
| 799 | void |
| 800 | globfree(glob_t *pglob) |
| 801 | { |
| 802 | size_t i; |
| 803 | char **pp; |
| 804 | |
| 805 | if (pglob->gl_pathv != NULL) { |
| 806 | pp = pglob->gl_pathv + pglob->gl_offs; |
| 807 | for (i = pglob->gl_pathc; i--; ++pp) |
| 808 | if (*pp) |
| 809 | free(*pp); |
| 810 | free(pglob->gl_pathv); |
| 811 | pglob->gl_pathv = NULL; |
| 812 | } |
| 813 | } |
| 814 | |
| 815 | static DIR * |
| 816 | g_opendir(Char *str, glob_t *pglob) |
| 817 | { |
| 818 | char buf[MAXPATHLEN]; |
| 819 | |
| 820 | if (!*str) |
| 821 | strcpy(buf, "."); |
| 822 | else { |
| 823 | if (g_Ctoc(str, buf, sizeof(buf))) |
| 824 | return (NULL); |
| 825 | } |
| 826 | |
| 827 | if (pglob->gl_flags & GLOB_ALTDIRFUNC) |
| 828 | return((*pglob->gl_opendir)(buf)); |
| 829 | |
| 830 | return(opendir(buf)); |
| 831 | } |
| 832 | |
| 833 | static int |
| 834 | g_lstat(Char *fn, struct stat *sb, glob_t *pglob) |
| 835 | { |
| 836 | char buf[MAXPATHLEN]; |
| 837 | |
| 838 | if (g_Ctoc(fn, buf, sizeof(buf))) { |
| 839 | errno = ENAMETOOLONG; |
| 840 | return (-1); |
| 841 | } |
| 842 | if (pglob->gl_flags & GLOB_ALTDIRFUNC) |
| 843 | return((*pglob->gl_lstat)(buf, sb)); |
| 844 | return(lstat(buf, sb)); |
| 845 | } |
| 846 | |
| 847 | static int |
| 848 | g_stat(Char *fn, struct stat *sb, glob_t *pglob) |
| 849 | { |
| 850 | char buf[MAXPATHLEN]; |
| 851 | |
| 852 | if (g_Ctoc(fn, buf, sizeof(buf))) { |
| 853 | errno = ENAMETOOLONG; |
| 854 | return (-1); |
| 855 | } |
| 856 | if (pglob->gl_flags & GLOB_ALTDIRFUNC) |
| 857 | return((*pglob->gl_stat)(buf, sb)); |
| 858 | return(stat(buf, sb)); |
| 859 | } |
| 860 | |
| 861 | static const Char * |
| 862 | g_strchr(const Char *str, wchar_t ch) |
| 863 | { |
| 864 | |
| 865 | do { |
| 866 | if (*str == ch) |
| 867 | return (str); |
| 868 | } while (*str++); |
| 869 | return (NULL); |
| 870 | } |
| 871 | |
| 872 | static int |
| 873 | g_Ctoc(const Char *str, char *buf, size_t len) |
| 874 | { |
| 875 | mbstate_t mbs; |
| 876 | size_t clen; |
| 877 | |
| 878 | memset(&mbs, 0, sizeof(mbs)); |
| 879 | while (len >= MB_CUR_MAX) { |
| 880 | clen = wcrtomb(buf, *str, &mbs); |
| 881 | if (clen == (size_t)-1) |
| 882 | return (1); |
| 883 | if (*str == L'\0') |
| 884 | return (0); |
| 885 | str++; |
| 886 | buf += clen; |
| 887 | len -= clen; |
| 888 | } |
| 889 | return (1); |
| 890 | } |
| 891 | |
| 892 | #ifdef DEBUG |
| 893 | static void |
| 894 | qprintf(const char *str, Char *s) |
| 895 | { |
| 896 | Char *p; |
| 897 | |
| 898 | (void)printf("%s:\n", str); |
| 899 | for (p = s; *p; p++) |
| 900 | (void)printf("%c", CHAR(*p)); |
| 901 | (void)printf("\n"); |
| 902 | for (p = s; *p; p++) |
| 903 | (void)printf("%c", *p & M_PROTECT ? '"' : ' '); |
| 904 | (void)printf("\n"); |
| 905 | for (p = s; *p; p++) |
| 906 | (void)printf("%c", ismeta(*p) ? '_' : ' '); |
| 907 | (void)printf("\n"); |
| 908 | } |
| 909 | #endif |