11573Srgrimes/* 21573Srgrimes * Copyright (c) 1989, 1993 31573Srgrimes * The Regents of the University of California. All rights reserved. 41573Srgrimes * 51573Srgrimes * This code is derived from software contributed to Berkeley by 61573Srgrimes * Guido van Rossum. 71573Srgrimes * 8227753Stheraven * Copyright (c) 2011 The FreeBSD Foundation 9227753Stheraven * All rights reserved. 10227753Stheraven * Portions of this software were developed by David Chisnall 11227753Stheraven * under sponsorship from the FreeBSD Foundation. 12227753Stheraven * 131573Srgrimes * Redistribution and use in source and binary forms, with or without 141573Srgrimes * modification, are permitted provided that the following conditions 151573Srgrimes * are met: 161573Srgrimes * 1. Redistributions of source code must retain the above copyright 171573Srgrimes * notice, this list of conditions and the following disclaimer. 181573Srgrimes * 2. Redistributions in binary form must reproduce the above copyright 191573Srgrimes * notice, this list of conditions and the following disclaimer in the 201573Srgrimes * documentation and/or other materials provided with the distribution. 211573Srgrimes * 4. Neither the name of the University nor the names of its contributors 221573Srgrimes * may be used to endorse or promote products derived from this software 231573Srgrimes * without specific prior written permission. 241573Srgrimes * 251573Srgrimes * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 261573Srgrimes * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 271573Srgrimes * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 281573Srgrimes * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 291573Srgrimes * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 301573Srgrimes * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 311573Srgrimes * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 321573Srgrimes * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 331573Srgrimes * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 341573Srgrimes * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 351573Srgrimes * SUCH DAMAGE. 361573Srgrimes */ 371573Srgrimes 381573Srgrimes#if defined(LIBC_SCCS) && !defined(lint) 391573Srgrimesstatic char sccsid[] = "@(#)glob.c 8.3 (Berkeley) 10/13/93"; 401573Srgrimes#endif /* LIBC_SCCS and not lint */ 4190045Sobrien#include <sys/cdefs.h> 4290045Sobrien__FBSDID("$FreeBSD$"); 431573Srgrimes 441573Srgrimes/* 451573Srgrimes * glob(3) -- a superset of the one defined in POSIX 1003.2. 461573Srgrimes * 471573Srgrimes * The [!...] convention to negate a range is supported (SysV, Posix, ksh). 481573Srgrimes * 491573Srgrimes * Optional extra services, controlled by flags not defined by POSIX: 501573Srgrimes * 511573Srgrimes * GLOB_QUOTE: 521573Srgrimes * Escaping convention: \ inhibits any special meaning the following 531573Srgrimes * character might have (except \ at end of string is retained). 541573Srgrimes * GLOB_MAGCHAR: 551573Srgrimes * Set in gl_flags if pattern contained a globbing character. 561573Srgrimes * GLOB_NOMAGIC: 571573Srgrimes * Same as GLOB_NOCHECK, but it will only append pattern if it did 581573Srgrimes * not contain any magic characters. [Used in csh style globbing] 591573Srgrimes * GLOB_ALTDIRFUNC: 601573Srgrimes * Use alternately specified directory access functions. 611573Srgrimes * GLOB_TILDE: 621573Srgrimes * expand ~user/foo to the /home/dir/of/user/foo 631573Srgrimes * GLOB_BRACE: 648870Srgrimes * expand {1,2}{a,b} to 1a 1b 2a 2b 651573Srgrimes * gl_matchc: 661573Srgrimes * Number of matches in the current invocation of glob. 671573Srgrimes */ 681573Srgrimes 69132817Stjr/* 70132817Stjr * Some notes on multibyte character support: 71132817Stjr * 1. Patterns with illegal byte sequences match nothing - even if 72132817Stjr * GLOB_NOCHECK is specified. 73132817Stjr * 2. Illegal byte sequences in filenames are handled by treating them as 74132817Stjr * single-byte characters with a value of the first byte of the sequence 75132817Stjr * cast to wchar_t. 76132817Stjr * 3. State-dependent encodings are not currently supported. 77132817Stjr */ 78132817Stjr 791573Srgrimes#include <sys/param.h> 801573Srgrimes#include <sys/stat.h> 811573Srgrimes 821573Srgrimes#include <ctype.h> 831573Srgrimes#include <dirent.h> 841573Srgrimes#include <errno.h> 851573Srgrimes#include <glob.h> 86132817Stjr#include <limits.h> 871573Srgrimes#include <pwd.h> 88132817Stjr#include <stdint.h> 891573Srgrimes#include <stdio.h> 901573Srgrimes#include <stdlib.h> 911573Srgrimes#include <string.h> 921573Srgrimes#include <unistd.h> 93132817Stjr#include <wchar.h> 941573Srgrimes 9519276Sache#include "collate.h" 9619276Sache 97243779Smarcel/* 98243779Smarcel * glob(3) expansion limits. Stop the expansion if any of these limits 99243779Smarcel * is reached. This caps the runtime in the face of DoS attacks. See 100243779Smarcel * also CVE-2010-2632 101243779Smarcel */ 102243779Smarcel#define GLOB_LIMIT_BRACE 128 /* number of brace calls */ 103243779Smarcel#define GLOB_LIMIT_PATH 65536 /* number of path elements */ 104243779Smarcel#define GLOB_LIMIT_READDIR 16384 /* number of readdirs */ 105243779Smarcel#define GLOB_LIMIT_STAT 1024 /* number of stat system calls */ 106243779Smarcel#define GLOB_LIMIT_STRING ARG_MAX /* maximum total size for paths */ 107243779Smarcel 108243779Smarcelstruct glob_limit { 109243779Smarcel size_t l_brace_cnt; 110243779Smarcel size_t l_path_lim; 111243779Smarcel size_t l_readdir_cnt; 112243779Smarcel size_t l_stat_cnt; 113243779Smarcel size_t l_string_cnt; 114243779Smarcel}; 115243779Smarcel 1161573Srgrimes#define DOLLAR '$' 1171573Srgrimes#define DOT '.' 1181573Srgrimes#define EOS '\0' 1191573Srgrimes#define LBRACKET '[' 1201573Srgrimes#define NOT '!' 1211573Srgrimes#define QUESTION '?' 1221573Srgrimes#define QUOTE '\\' 1231573Srgrimes#define RANGE '-' 1241573Srgrimes#define RBRACKET ']' 1251573Srgrimes#define SEP '/' 1261573Srgrimes#define STAR '*' 1271573Srgrimes#define TILDE '~' 1281573Srgrimes#define UNDERSCORE '_' 1291573Srgrimes#define LBRACE '{' 1301573Srgrimes#define RBRACE '}' 1311573Srgrimes#define SLASH '/' 1321573Srgrimes#define COMMA ',' 1331573Srgrimes 1341573Srgrimes#ifndef DEBUG 1351573Srgrimes 136132817Stjr#define M_QUOTE 0x8000000000ULL 137132817Stjr#define M_PROTECT 0x4000000000ULL 138132817Stjr#define M_MASK 0xffffffffffULL 139132817Stjr#define M_CHAR 0x00ffffffffULL 1401573Srgrimes 141132817Stjrtypedef uint_fast64_t Char; 1421573Srgrimes 1431573Srgrimes#else 1441573Srgrimes 1451573Srgrimes#define M_QUOTE 0x80 1461573Srgrimes#define M_PROTECT 0x40 1471573Srgrimes#define M_MASK 0xff 148132817Stjr#define M_CHAR 0x7f 1491573Srgrimes 1501573Srgrimestypedef char Char; 1511573Srgrimes 1521573Srgrimes#endif 1531573Srgrimes 1541573Srgrimes 155132817Stjr#define CHAR(c) ((Char)((c)&M_CHAR)) 1561573Srgrimes#define META(c) ((Char)((c)|M_QUOTE)) 1571573Srgrimes#define M_ALL META('*') 1581573Srgrimes#define M_END META(']') 1591573Srgrimes#define M_NOT META('!') 1601573Srgrimes#define M_ONE META('?') 1611573Srgrimes#define M_RNG META('-') 1621573Srgrimes#define M_SET META('[') 1631573Srgrimes#define ismeta(c) (((c)&M_QUOTE) != 0) 1641573Srgrimes 1651573Srgrimes 16690045Sobrienstatic int compare(const void *, const void *); 167158812Sachestatic int g_Ctoc(const Char *, char *, size_t); 16890045Sobrienstatic int g_lstat(Char *, struct stat *, glob_t *); 16990045Sobrienstatic DIR *g_opendir(Char *, glob_t *); 170180021Smtmstatic const Char *g_strchr(const Char *, wchar_t); 1711573Srgrimes#ifdef notdef 17290045Sobrienstatic Char *g_strcat(Char *, const Char *); 1731573Srgrimes#endif 17490045Sobrienstatic int g_stat(Char *, struct stat *, glob_t *); 175243779Smarcelstatic int glob0(const Char *, glob_t *, struct glob_limit *); 176243779Smarcelstatic int glob1(Char *, glob_t *, struct glob_limit *); 177243779Smarcelstatic int glob2(Char *, Char *, Char *, Char *, glob_t *, 178243779Smarcel struct glob_limit *); 179243779Smarcelstatic int glob3(Char *, Char *, Char *, Char *, Char *, glob_t *, 180243779Smarcel struct glob_limit *); 181243779Smarcelstatic int globextend(const Char *, glob_t *, struct glob_limit *); 182243779Smarcelstatic const Char * 18390045Sobrien globtilde(const Char *, Char *, size_t, glob_t *); 184243779Smarcelstatic int globexp1(const Char *, glob_t *, struct glob_limit *); 185243779Smarcelstatic int globexp2(const Char *, const Char *, glob_t *, int *, 186243779Smarcel struct glob_limit *); 18790045Sobrienstatic int match(Char *, Char *, Char *); 1881573Srgrimes#ifdef DEBUG 18990045Sobrienstatic void qprintf(const char *, Char *); 1901573Srgrimes#endif 1911573Srgrimes 1921573Srgrimesint 193228754Seadlerglob(const char * __restrict pattern, int flags, 194228754Seadler int (*errfunc)(const char *, int), glob_t * __restrict pglob) 1951573Srgrimes{ 196243779Smarcel struct glob_limit limit = { 0, 0, 0, 0, 0 }; 197159294Sdelphij const char *patnext; 198132817Stjr Char *bufnext, *bufend, patbuf[MAXPATHLEN], prot; 199132817Stjr mbstate_t mbs; 200132817Stjr wchar_t wc; 201132817Stjr size_t clen; 2021573Srgrimes 203159294Sdelphij patnext = pattern; 2041573Srgrimes if (!(flags & GLOB_APPEND)) { 2051573Srgrimes pglob->gl_pathc = 0; 2061573Srgrimes pglob->gl_pathv = NULL; 2071573Srgrimes if (!(flags & GLOB_DOOFFS)) 2081573Srgrimes pglob->gl_offs = 0; 2091573Srgrimes } 21080525Smikeh if (flags & GLOB_LIMIT) { 211243779Smarcel limit.l_path_lim = pglob->gl_matchc; 212243779Smarcel if (limit.l_path_lim == 0) 213243779Smarcel limit.l_path_lim = GLOB_LIMIT_PATH; 214243779Smarcel } 2151573Srgrimes pglob->gl_flags = flags & ~GLOB_MAGCHAR; 2161573Srgrimes pglob->gl_errfunc = errfunc; 2171573Srgrimes pglob->gl_matchc = 0; 2181573Srgrimes 2191573Srgrimes bufnext = patbuf; 22074963Speter bufend = bufnext + MAXPATHLEN - 1; 221132817Stjr if (flags & GLOB_NOESCAPE) { 222132817Stjr memset(&mbs, 0, sizeof(mbs)); 223132817Stjr while (bufend - bufnext >= MB_CUR_MAX) { 224132817Stjr clen = mbrtowc(&wc, patnext, MB_LEN_MAX, &mbs); 225132817Stjr if (clen == (size_t)-1 || clen == (size_t)-2) 226132817Stjr return (GLOB_NOMATCH); 227132817Stjr else if (clen == 0) 228132817Stjr break; 229132817Stjr *bufnext++ = wc; 230132817Stjr patnext += clen; 231132817Stjr } 232132817Stjr } else { 2331573Srgrimes /* Protect the quoted characters. */ 234132817Stjr memset(&mbs, 0, sizeof(mbs)); 235132817Stjr while (bufend - bufnext >= MB_CUR_MAX) { 236132817Stjr if (*patnext == QUOTE) { 237132817Stjr if (*++patnext == EOS) { 238132817Stjr *bufnext++ = QUOTE | M_PROTECT; 239132817Stjr continue; 2401573Srgrimes } 241132817Stjr prot = M_PROTECT; 242132817Stjr } else 243132817Stjr prot = 0; 244132817Stjr clen = mbrtowc(&wc, patnext, MB_LEN_MAX, &mbs); 245132817Stjr if (clen == (size_t)-1 || clen == (size_t)-2) 246132817Stjr return (GLOB_NOMATCH); 247132817Stjr else if (clen == 0) 248132817Stjr break; 249132817Stjr *bufnext++ = wc | prot; 250132817Stjr patnext += clen; 251132817Stjr } 2521573Srgrimes } 2531573Srgrimes *bufnext = EOS; 2541573Srgrimes 2551573Srgrimes if (flags & GLOB_BRACE) 256228755Seadler return (globexp1(patbuf, pglob, &limit)); 2571573Srgrimes else 258228755Seadler return (glob0(patbuf, pglob, &limit)); 2591573Srgrimes} 2601573Srgrimes 2611573Srgrimes/* 2621573Srgrimes * Expand recursively a glob {} pattern. When there is no more expansion 2631573Srgrimes * invoke the standard globbing routine to glob the rest of the magic 2641573Srgrimes * characters 2651573Srgrimes */ 26674963Speterstatic int 267243779Smarcelglobexp1(const Char *pattern, glob_t *pglob, struct glob_limit *limit) 2681573Srgrimes{ 2691573Srgrimes const Char* ptr = pattern; 2701573Srgrimes int rv; 2711573Srgrimes 272243779Smarcel if ((pglob->gl_flags & GLOB_LIMIT) && 273243779Smarcel limit->l_brace_cnt++ >= GLOB_LIMIT_BRACE) { 274243779Smarcel errno = 0; 275243779Smarcel return (GLOB_NOSPACE); 276243779Smarcel } 277243779Smarcel 2781573Srgrimes /* Protect a single {}, for find(1), like csh */ 2791573Srgrimes if (pattern[0] == LBRACE && pattern[1] == RBRACE && pattern[2] == EOS) 28074469Sjlemon return glob0(pattern, pglob, limit); 2811573Srgrimes 282180021Smtm while ((ptr = g_strchr(ptr, LBRACE)) != NULL) 28374469Sjlemon if (!globexp2(ptr, pattern, pglob, &rv, limit)) 2841573Srgrimes return rv; 2851573Srgrimes 28674469Sjlemon return glob0(pattern, pglob, limit); 2871573Srgrimes} 2881573Srgrimes 2891573Srgrimes 2901573Srgrimes/* 2911573Srgrimes * Recursive brace globbing helper. Tries to expand a single brace. 2921573Srgrimes * If it succeeds then it invokes globexp1 with the new pattern. 2931573Srgrimes * If it fails then it tries to glob the rest of the pattern and returns. 2941573Srgrimes */ 29574963Speterstatic int 296243779Smarcelglobexp2(const Char *ptr, const Char *pattern, glob_t *pglob, int *rv, 297243779Smarcel struct glob_limit *limit) 2981573Srgrimes{ 2991573Srgrimes int i; 3001573Srgrimes Char *lm, *ls; 301150137Sache const Char *pe, *pm, *pm1, *pl; 30274963Speter Char patbuf[MAXPATHLEN]; 3031573Srgrimes 3041573Srgrimes /* copy part up to the brace */ 3051573Srgrimes for (lm = patbuf, pm = pattern; pm != ptr; *lm++ = *pm++) 3061573Srgrimes continue; 30774963Speter *lm = EOS; 3081573Srgrimes ls = lm; 3091573Srgrimes 3101573Srgrimes /* Find the balanced brace */ 3111573Srgrimes for (i = 0, pe = ++ptr; *pe; pe++) 3121573Srgrimes if (*pe == LBRACKET) { 3131573Srgrimes /* Ignore everything between [] */ 3141573Srgrimes for (pm = pe++; *pe != RBRACKET && *pe != EOS; pe++) 3151573Srgrimes continue; 3161573Srgrimes if (*pe == EOS) { 3178870Srgrimes /* 3181573Srgrimes * We could not find a matching RBRACKET. 3191573Srgrimes * Ignore and just look for RBRACE 3201573Srgrimes */ 3211573Srgrimes pe = pm; 3221573Srgrimes } 3231573Srgrimes } 3241573Srgrimes else if (*pe == LBRACE) 3251573Srgrimes i++; 3261573Srgrimes else if (*pe == RBRACE) { 3271573Srgrimes if (i == 0) 3281573Srgrimes break; 3291573Srgrimes i--; 3301573Srgrimes } 3311573Srgrimes 3321573Srgrimes /* Non matching braces; just glob the pattern */ 3331573Srgrimes if (i != 0 || *pe == EOS) { 33474469Sjlemon *rv = glob0(patbuf, pglob, limit); 335228755Seadler return (0); 3361573Srgrimes } 3371573Srgrimes 3381573Srgrimes for (i = 0, pl = pm = ptr; pm <= pe; pm++) 3391573Srgrimes switch (*pm) { 3401573Srgrimes case LBRACKET: 3411573Srgrimes /* Ignore everything between [] */ 342150137Sache for (pm1 = pm++; *pm != RBRACKET && *pm != EOS; pm++) 3431573Srgrimes continue; 3441573Srgrimes if (*pm == EOS) { 3458870Srgrimes /* 3461573Srgrimes * We could not find a matching RBRACKET. 3471573Srgrimes * Ignore and just look for RBRACE 3481573Srgrimes */ 349150137Sache pm = pm1; 3501573Srgrimes } 3511573Srgrimes break; 3521573Srgrimes 3531573Srgrimes case LBRACE: 3541573Srgrimes i++; 3551573Srgrimes break; 3561573Srgrimes 3571573Srgrimes case RBRACE: 3581573Srgrimes if (i) { 3591573Srgrimes i--; 3601573Srgrimes break; 3611573Srgrimes } 3621573Srgrimes /* FALLTHROUGH */ 3631573Srgrimes case COMMA: 3641573Srgrimes if (i && *pm == COMMA) 3651573Srgrimes break; 3661573Srgrimes else { 3671573Srgrimes /* Append the current string */ 3681573Srgrimes for (lm = ls; (pl < pm); *lm++ = *pl++) 3691573Srgrimes continue; 3708870Srgrimes /* 3711573Srgrimes * Append the rest of the pattern after the 3721573Srgrimes * closing brace 3731573Srgrimes */ 3741573Srgrimes for (pl = pe + 1; (*lm++ = *pl++) != EOS;) 3751573Srgrimes continue; 3761573Srgrimes 3771573Srgrimes /* Expand the current pattern */ 3781573Srgrimes#ifdef DEBUG 3791573Srgrimes qprintf("globexp2:", patbuf); 3801573Srgrimes#endif 38174469Sjlemon *rv = globexp1(patbuf, pglob, limit); 3821573Srgrimes 3831573Srgrimes /* move after the comma, to the next string */ 3841573Srgrimes pl = pm + 1; 3851573Srgrimes } 3861573Srgrimes break; 3871573Srgrimes 3881573Srgrimes default: 3891573Srgrimes break; 3901573Srgrimes } 3911573Srgrimes *rv = 0; 392228755Seadler return (0); 3931573Srgrimes} 3941573Srgrimes 3951573Srgrimes 3961573Srgrimes 3971573Srgrimes/* 3981573Srgrimes * expand tilde from the passwd file. 3991573Srgrimes */ 4001573Srgrimesstatic const Char * 401159294Sdelphijglobtilde(const Char *pattern, Char *patbuf, size_t patbuf_len, glob_t *pglob) 4021573Srgrimes{ 4031573Srgrimes struct passwd *pwd; 4041573Srgrimes char *h; 4051573Srgrimes const Char *p; 40624158Simp Char *b, *eb; 4071573Srgrimes 4081573Srgrimes if (*pattern != TILDE || !(pglob->gl_flags & GLOB_TILDE)) 409228755Seadler return (pattern); 4101573Srgrimes 41124158Simp /* 41224158Simp * Copy up to the end of the string or / 41324158Simp */ 41424158Simp eb = &patbuf[patbuf_len - 1]; 41524158Simp for (p = pattern + 1, h = (char *) patbuf; 41624158Simp h < (char *)eb && *p && *p != SLASH; *h++ = *p++) 4171573Srgrimes continue; 4181573Srgrimes 4191573Srgrimes *h = EOS; 4201573Srgrimes 4211573Srgrimes if (((char *) patbuf)[0] == EOS) { 4228870Srgrimes /* 42328820Simp * handle a plain ~ or ~/ by expanding $HOME first (iff 42428820Simp * we're not running setuid or setgid) and then trying 42528820Simp * the password file 4261573Srgrimes */ 427121667Stjr if (issetugid() != 0 || 42833664Sjb (h = getenv("HOME")) == NULL) { 42928836Sache if (((h = getlogin()) != NULL && 43028836Sache (pwd = getpwnam(h)) != NULL) || 43128836Sache (pwd = getpwuid(getuid())) != NULL) 43228836Sache h = pwd->pw_dir; 43328836Sache else 434228755Seadler return (pattern); 4351573Srgrimes } 4361573Srgrimes } 4371573Srgrimes else { 4381573Srgrimes /* 4391573Srgrimes * Expand a ~user 4401573Srgrimes */ 4411573Srgrimes if ((pwd = getpwnam((char*) patbuf)) == NULL) 442228755Seadler return (pattern); 4431573Srgrimes else 4441573Srgrimes h = pwd->pw_dir; 4451573Srgrimes } 4461573Srgrimes 4471573Srgrimes /* Copy the home directory */ 44824158Simp for (b = patbuf; b < eb && *h; *b++ = *h++) 4491573Srgrimes continue; 4508870Srgrimes 4511573Srgrimes /* Append the rest of the pattern */ 45224158Simp while (b < eb && (*b++ = *p++) != EOS) 4531573Srgrimes continue; 45424158Simp *b = EOS; 4551573Srgrimes 456228755Seadler return (patbuf); 4571573Srgrimes} 4581573Srgrimes 4598870Srgrimes 4601573Srgrimes/* 4611573Srgrimes * The main glob() routine: compiles the pattern (optionally processing 4621573Srgrimes * quotes), calls glob1() to do the real pattern matching, and finally 4631573Srgrimes * sorts the list (unless unsorted operation is requested). Returns 0 464100217Smikeh * if things went well, nonzero if errors occurred. 4651573Srgrimes */ 4661573Srgrimesstatic int 467243779Smarcelglob0(const Char *pattern, glob_t *pglob, struct glob_limit *limit) 4681573Srgrimes{ 4691573Srgrimes const Char *qpatnext; 470207981Sgordon int err; 471158812Sache size_t oldpathc; 472207981Sgordon Char *bufnext, c, patbuf[MAXPATHLEN]; 4731573Srgrimes 47474963Speter qpatnext = globtilde(pattern, patbuf, MAXPATHLEN, pglob); 4751573Srgrimes oldpathc = pglob->gl_pathc; 4761573Srgrimes bufnext = patbuf; 4771573Srgrimes 4781573Srgrimes /* We don't need to check for buffer overflow any more. */ 4791573Srgrimes while ((c = *qpatnext++) != EOS) { 4801573Srgrimes switch (c) { 4811573Srgrimes case LBRACKET: 4821573Srgrimes c = *qpatnext; 4831573Srgrimes if (c == NOT) 4841573Srgrimes ++qpatnext; 4851573Srgrimes if (*qpatnext == EOS || 486180021Smtm g_strchr(qpatnext+1, RBRACKET) == NULL) { 4871573Srgrimes *bufnext++ = LBRACKET; 4881573Srgrimes if (c == NOT) 4891573Srgrimes --qpatnext; 4901573Srgrimes break; 4911573Srgrimes } 4921573Srgrimes *bufnext++ = M_SET; 4931573Srgrimes if (c == NOT) 4941573Srgrimes *bufnext++ = M_NOT; 4951573Srgrimes c = *qpatnext++; 4961573Srgrimes do { 4971573Srgrimes *bufnext++ = CHAR(c); 4981573Srgrimes if (*qpatnext == RANGE && 4991573Srgrimes (c = qpatnext[1]) != RBRACKET) { 5001573Srgrimes *bufnext++ = M_RNG; 5011573Srgrimes *bufnext++ = CHAR(c); 5021573Srgrimes qpatnext += 2; 5031573Srgrimes } 5041573Srgrimes } while ((c = *qpatnext++) != RBRACKET); 5051573Srgrimes pglob->gl_flags |= GLOB_MAGCHAR; 5061573Srgrimes *bufnext++ = M_END; 5071573Srgrimes break; 5081573Srgrimes case QUESTION: 5091573Srgrimes pglob->gl_flags |= GLOB_MAGCHAR; 5101573Srgrimes *bufnext++ = M_ONE; 5111573Srgrimes break; 5121573Srgrimes case STAR: 5131573Srgrimes pglob->gl_flags |= GLOB_MAGCHAR; 5148870Srgrimes /* collapse adjacent stars to one, 5151573Srgrimes * to avoid exponential behavior 5161573Srgrimes */ 5171573Srgrimes if (bufnext == patbuf || bufnext[-1] != M_ALL) 5181573Srgrimes *bufnext++ = M_ALL; 5191573Srgrimes break; 5201573Srgrimes default: 5211573Srgrimes *bufnext++ = CHAR(c); 5221573Srgrimes break; 5231573Srgrimes } 5241573Srgrimes } 5251573Srgrimes *bufnext = EOS; 5261573Srgrimes#ifdef DEBUG 5271573Srgrimes qprintf("glob0:", patbuf); 5281573Srgrimes#endif 5291573Srgrimes 53074469Sjlemon if ((err = glob1(patbuf, pglob, limit)) != 0) 5311573Srgrimes return(err); 5321573Srgrimes 5331573Srgrimes /* 5348870Srgrimes * If there was no match we are going to append the pattern 5351573Srgrimes * if GLOB_NOCHECK was specified or if GLOB_NOMAGIC was specified 5361573Srgrimes * and the pattern did not contain any magic characters 5371573Srgrimes * GLOB_NOMAGIC is there just for compatibility with csh. 5381573Srgrimes */ 539100217Smikeh if (pglob->gl_pathc == oldpathc) { 540100217Smikeh if (((pglob->gl_flags & GLOB_NOCHECK) || 541100217Smikeh ((pglob->gl_flags & GLOB_NOMAGIC) && 542100217Smikeh !(pglob->gl_flags & GLOB_MAGCHAR)))) 543228755Seadler return (globextend(pattern, pglob, limit)); 544100217Smikeh else 545228755Seadler return (GLOB_NOMATCH); 546100217Smikeh } 547100217Smikeh if (!(pglob->gl_flags & GLOB_NOSORT)) 5481573Srgrimes qsort(pglob->gl_pathv + pglob->gl_offs + oldpathc, 5491573Srgrimes pglob->gl_pathc - oldpathc, sizeof(char *), compare); 550228755Seadler return (0); 5511573Srgrimes} 5521573Srgrimes 5531573Srgrimesstatic int 554159294Sdelphijcompare(const void *p, const void *q) 5551573Srgrimes{ 556228755Seadler return (strcmp(*(char **)p, *(char **)q)); 5571573Srgrimes} 5581573Srgrimes 5591573Srgrimesstatic int 560243779Smarcelglob1(Char *pattern, glob_t *pglob, struct glob_limit *limit) 5611573Srgrimes{ 56274963Speter Char pathbuf[MAXPATHLEN]; 5631573Srgrimes 5641573Srgrimes /* A null pathname is invalid -- POSIX 1003.1 sect. 2.4. */ 5651573Srgrimes if (*pattern == EOS) 566228755Seadler return (0); 567228755Seadler return (glob2(pathbuf, pathbuf, pathbuf + MAXPATHLEN - 1, 56874963Speter pattern, pglob, limit)); 5691573Srgrimes} 5701573Srgrimes 5711573Srgrimes/* 5721573Srgrimes * The functions glob2 and glob3 are mutually recursive; there is one level 5731573Srgrimes * of recursion for each segment in the pattern that contains one or more 5741573Srgrimes * meta characters. 5751573Srgrimes */ 5761573Srgrimesstatic int 577159294Sdelphijglob2(Char *pathbuf, Char *pathend, Char *pathend_last, Char *pattern, 578243779Smarcel glob_t *pglob, struct glob_limit *limit) 5791573Srgrimes{ 5801573Srgrimes struct stat sb; 5811573Srgrimes Char *p, *q; 5821573Srgrimes int anymeta; 5831573Srgrimes 5841573Srgrimes /* 5851573Srgrimes * Loop over pattern segments until end of pattern or until 5861573Srgrimes * segment with meta character found. 5871573Srgrimes */ 5881573Srgrimes for (anymeta = 0;;) { 5891573Srgrimes if (*pattern == EOS) { /* End of pattern? */ 5901573Srgrimes *pathend = EOS; 5911573Srgrimes if (g_lstat(pathbuf, &sb, pglob)) 592228755Seadler return (0); 5938870Srgrimes 594243779Smarcel if ((pglob->gl_flags & GLOB_LIMIT) && 595243779Smarcel limit->l_stat_cnt++ >= GLOB_LIMIT_STAT) { 596243779Smarcel errno = 0; 597243779Smarcel if (pathend + 1 > pathend_last) 598243779Smarcel return (GLOB_ABORTED); 599243779Smarcel *pathend++ = SEP; 600243779Smarcel *pathend = EOS; 601243779Smarcel return (GLOB_NOSPACE); 602243779Smarcel } 6031573Srgrimes if (((pglob->gl_flags & GLOB_MARK) && 6041573Srgrimes pathend[-1] != SEP) && (S_ISDIR(sb.st_mode) 6051573Srgrimes || (S_ISLNK(sb.st_mode) && 6061573Srgrimes (g_stat(pathbuf, &sb, pglob) == 0) && 6071573Srgrimes S_ISDIR(sb.st_mode)))) { 60874963Speter if (pathend + 1 > pathend_last) 609100217Smikeh return (GLOB_ABORTED); 6101573Srgrimes *pathend++ = SEP; 6111573Srgrimes *pathend = EOS; 6121573Srgrimes } 6131573Srgrimes ++pglob->gl_matchc; 614228755Seadler return (globextend(pathbuf, pglob, limit)); 6151573Srgrimes } 6161573Srgrimes 6171573Srgrimes /* Find end of next segment, copy tentatively to pathend. */ 6181573Srgrimes q = pathend; 6191573Srgrimes p = pattern; 6201573Srgrimes while (*p != EOS && *p != SEP) { 6211573Srgrimes if (ismeta(*p)) 6221573Srgrimes anymeta = 1; 62374963Speter if (q + 1 > pathend_last) 624100217Smikeh return (GLOB_ABORTED); 6251573Srgrimes *q++ = *p++; 6261573Srgrimes } 6271573Srgrimes 6281573Srgrimes if (!anymeta) { /* No expansion, do next segment. */ 6291573Srgrimes pathend = q; 6301573Srgrimes pattern = p; 63174963Speter while (*pattern == SEP) { 63274963Speter if (pathend + 1 > pathend_last) 633100217Smikeh return (GLOB_ABORTED); 6341573Srgrimes *pathend++ = *pattern++; 63574963Speter } 6361573Srgrimes } else /* Need expansion, recurse. */ 637228755Seadler return (glob3(pathbuf, pathend, pathend_last, pattern, 638228755Seadler p, pglob, limit)); 6391573Srgrimes } 6401573Srgrimes /* NOTREACHED */ 6411573Srgrimes} 6421573Srgrimes 6431573Srgrimesstatic int 644159294Sdelphijglob3(Char *pathbuf, Char *pathend, Char *pathend_last, 645159294Sdelphij Char *pattern, Char *restpattern, 646243779Smarcel glob_t *pglob, struct glob_limit *limit) 6471573Srgrimes{ 64890045Sobrien struct dirent *dp; 6491573Srgrimes DIR *dirp; 6501573Srgrimes int err; 6511573Srgrimes char buf[MAXPATHLEN]; 6521573Srgrimes 6531573Srgrimes /* 6541573Srgrimes * The readdirfunc declaration can't be prototyped, because it is 6551573Srgrimes * assigned, below, to two functions which are prototyped in glob.h 6561573Srgrimes * and dirent.h as taking pointers to differently typed opaque 6571573Srgrimes * structures. 6581573Srgrimes */ 6591573Srgrimes struct dirent *(*readdirfunc)(); 6601573Srgrimes 66174963Speter if (pathend > pathend_last) 662100217Smikeh return (GLOB_ABORTED); 6631573Srgrimes *pathend = EOS; 6641573Srgrimes errno = 0; 6658870Srgrimes 6661573Srgrimes if ((dirp = g_opendir(pathbuf, pglob)) == NULL) { 6671573Srgrimes /* TODO: don't call for ENOENT or ENOTDIR? */ 6681573Srgrimes if (pglob->gl_errfunc) { 66974921Speter if (g_Ctoc(pathbuf, buf, sizeof(buf))) 670100217Smikeh return (GLOB_ABORTED); 6711573Srgrimes if (pglob->gl_errfunc(buf, errno) || 6721573Srgrimes pglob->gl_flags & GLOB_ERR) 673100217Smikeh return (GLOB_ABORTED); 6741573Srgrimes } 675228755Seadler return (0); 6761573Srgrimes } 6771573Srgrimes 6781573Srgrimes err = 0; 6791573Srgrimes 6801573Srgrimes /* Search directory for matching names. */ 6811573Srgrimes if (pglob->gl_flags & GLOB_ALTDIRFUNC) 6821573Srgrimes readdirfunc = pglob->gl_readdir; 6831573Srgrimes else 6841573Srgrimes readdirfunc = readdir; 6851573Srgrimes while ((dp = (*readdirfunc)(dirp))) { 686159294Sdelphij char *sc; 68790045Sobrien Char *dc; 688132817Stjr wchar_t wc; 689132817Stjr size_t clen; 690132817Stjr mbstate_t mbs; 6911573Srgrimes 692243779Smarcel if ((pglob->gl_flags & GLOB_LIMIT) && 693243779Smarcel limit->l_readdir_cnt++ >= GLOB_LIMIT_READDIR) { 694243779Smarcel errno = 0; 695243779Smarcel if (pathend + 1 > pathend_last) 696243779Smarcel err = GLOB_ABORTED; 697243779Smarcel else { 698243779Smarcel *pathend++ = SEP; 699243779Smarcel *pathend = EOS; 700243779Smarcel err = GLOB_NOSPACE; 701243779Smarcel } 702243779Smarcel break; 703243779Smarcel } 704243779Smarcel 7051573Srgrimes /* Initial DOT must be matched literally. */ 7061573Srgrimes if (dp->d_name[0] == DOT && *pattern != DOT) 7071573Srgrimes continue; 708132817Stjr memset(&mbs, 0, sizeof(mbs)); 70974963Speter dc = pathend; 710159294Sdelphij sc = dp->d_name; 711132817Stjr while (dc < pathend_last) { 712132817Stjr clen = mbrtowc(&wc, sc, MB_LEN_MAX, &mbs); 713132817Stjr if (clen == (size_t)-1 || clen == (size_t)-2) { 714132817Stjr wc = *sc; 715132817Stjr clen = 1; 716132817Stjr memset(&mbs, 0, sizeof(mbs)); 717132817Stjr } 718132817Stjr if ((*dc++ = wc) == EOS) 719132817Stjr break; 720132817Stjr sc += clen; 721132817Stjr } 7221573Srgrimes if (!match(pathend, pattern, restpattern)) { 7231573Srgrimes *pathend = EOS; 7241573Srgrimes continue; 7251573Srgrimes } 72674963Speter err = glob2(pathbuf, --dc, pathend_last, restpattern, 72774963Speter pglob, limit); 7281573Srgrimes if (err) 7291573Srgrimes break; 7301573Srgrimes } 7311573Srgrimes 7321573Srgrimes if (pglob->gl_flags & GLOB_ALTDIRFUNC) 7331573Srgrimes (*pglob->gl_closedir)(dirp); 7341573Srgrimes else 7351573Srgrimes closedir(dirp); 736228755Seadler return (err); 7371573Srgrimes} 7381573Srgrimes 7391573Srgrimes 7401573Srgrimes/* 741249381Semaste * Extend the gl_pathv member of a glob_t structure to accommodate a new item, 7421573Srgrimes * add the new item, and update gl_pathc. 7431573Srgrimes * 7441573Srgrimes * This assumes the BSD realloc, which only copies the block when its size 7451573Srgrimes * crosses a power-of-two boundary; for v7 realloc, this would cause quadratic 7461573Srgrimes * behavior. 7471573Srgrimes * 7481573Srgrimes * Return 0 if new item added, error code if memory couldn't be allocated. 7491573Srgrimes * 7501573Srgrimes * Invariant of the glob_t structure: 7511573Srgrimes * Either gl_pathc is zero and gl_pathv is NULL; or gl_pathc > 0 and 7521573Srgrimes * gl_pathv points to (gl_offs + gl_pathc + 1) items. 7531573Srgrimes */ 7541573Srgrimesstatic int 755243779Smarcelglobextend(const Char *path, glob_t *pglob, struct glob_limit *limit) 7561573Srgrimes{ 75790045Sobrien char **pathv; 758158812Sache size_t i, newsize, len; 7591573Srgrimes char *copy; 7601573Srgrimes const Char *p; 7611573Srgrimes 762243779Smarcel if ((pglob->gl_flags & GLOB_LIMIT) && 763243779Smarcel pglob->gl_matchc > limit->l_path_lim) { 76480525Smikeh errno = 0; 76580525Smikeh return (GLOB_NOSPACE); 76680525Smikeh } 76774307Sjlemon 7681573Srgrimes newsize = sizeof(*pathv) * (2 + pglob->gl_pathc + pglob->gl_offs); 769243759Smarcel /* realloc(NULL, newsize) is equivalent to malloc(newsize). */ 770243759Smarcel pathv = realloc((void *)pglob->gl_pathv, newsize); 771243758Smarcel if (pathv == NULL) 772228755Seadler return (GLOB_NOSPACE); 7731573Srgrimes 7741573Srgrimes if (pglob->gl_pathv == NULL && pglob->gl_offs > 0) { 7751573Srgrimes /* first time around -- clear initial gl_offs items */ 7761573Srgrimes pathv += pglob->gl_offs; 777158812Sache for (i = pglob->gl_offs + 1; --i > 0; ) 7781573Srgrimes *--pathv = NULL; 7791573Srgrimes } 7801573Srgrimes pglob->gl_pathv = pathv; 7811573Srgrimes 7821573Srgrimes for (p = path; *p++;) 7831573Srgrimes continue; 784132817Stjr len = MB_CUR_MAX * (size_t)(p - path); /* XXX overallocation */ 785243779Smarcel limit->l_string_cnt += len; 786243779Smarcel if ((pglob->gl_flags & GLOB_LIMIT) && 787243779Smarcel limit->l_string_cnt >= GLOB_LIMIT_STRING) { 788243779Smarcel errno = 0; 789243779Smarcel return (GLOB_NOSPACE); 790243779Smarcel } 79174921Speter if ((copy = malloc(len)) != NULL) { 79274921Speter if (g_Ctoc(path, copy, len)) { 79374918Speter free(copy); 79474918Speter return (GLOB_NOSPACE); 79574918Speter } 7961573Srgrimes pathv[pglob->gl_offs + pglob->gl_pathc++] = copy; 7971573Srgrimes } 7981573Srgrimes pathv[pglob->gl_offs + pglob->gl_pathc] = NULL; 799228755Seadler return (copy == NULL ? GLOB_NOSPACE : 0); 8001573Srgrimes} 8011573Srgrimes 8021573Srgrimes/* 8031573Srgrimes * pattern matching function for filenames. Each occurrence of the * 8041573Srgrimes * pattern causes a recursion level. 8051573Srgrimes */ 8061573Srgrimesstatic int 807159294Sdelphijmatch(Char *name, Char *pat, Char *patend) 8081573Srgrimes{ 8091573Srgrimes int ok, negate_range; 8101573Srgrimes Char c, k; 811227753Stheraven struct xlocale_collate *table = 812227753Stheraven (struct xlocale_collate*)__get_locale()->components[XLC_COLLATE]; 8131573Srgrimes 8141573Srgrimes while (pat < patend) { 8151573Srgrimes c = *pat++; 8161573Srgrimes switch (c & M_MASK) { 8171573Srgrimes case M_ALL: 8181573Srgrimes if (pat == patend) 819228755Seadler return (1); 8208870Srgrimes do 8211573Srgrimes if (match(name, pat, patend)) 822228755Seadler return (1); 8231573Srgrimes while (*name++ != EOS); 824228755Seadler return (0); 8251573Srgrimes case M_ONE: 8261573Srgrimes if (*name++ == EOS) 827228755Seadler return (0); 8281573Srgrimes break; 8291573Srgrimes case M_SET: 8301573Srgrimes ok = 0; 8311573Srgrimes if ((k = *name++) == EOS) 832228755Seadler return (0); 8331573Srgrimes if ((negate_range = ((*pat & M_MASK) == M_NOT)) != EOS) 8341573Srgrimes ++pat; 8351573Srgrimes while (((c = *pat++) & M_MASK) != M_END) 8361573Srgrimes if ((*pat & M_MASK) == M_RNG) { 837227753Stheraven if (table->__collate_load_error ? 83824633Sache CHAR(c) <= CHAR(k) && CHAR(k) <= CHAR(pat[1]) : 839227753Stheraven __collate_range_cmp(table, CHAR(c), CHAR(k)) <= 0 840227753Stheraven && __collate_range_cmp(table, CHAR(k), CHAR(pat[1])) <= 0 84117528Sache ) 8421573Srgrimes ok = 1; 8431573Srgrimes pat += 2; 8441573Srgrimes } else if (c == k) 8451573Srgrimes ok = 1; 8461573Srgrimes if (ok == negate_range) 847228755Seadler return (0); 8481573Srgrimes break; 8491573Srgrimes default: 8501573Srgrimes if (*name++ != c) 851228755Seadler return (0); 8521573Srgrimes break; 8531573Srgrimes } 8541573Srgrimes } 855228755Seadler return (*name == EOS); 8561573Srgrimes} 8571573Srgrimes 8581573Srgrimes/* Free allocated data belonging to a glob_t structure. */ 8591573Srgrimesvoid 860159294Sdelphijglobfree(glob_t *pglob) 8611573Srgrimes{ 862158812Sache size_t i; 86390045Sobrien char **pp; 8641573Srgrimes 8651573Srgrimes if (pglob->gl_pathv != NULL) { 8661573Srgrimes pp = pglob->gl_pathv + pglob->gl_offs; 8671573Srgrimes for (i = pglob->gl_pathc; i--; ++pp) 8681573Srgrimes if (*pp) 8691573Srgrimes free(*pp); 8701573Srgrimes free(pglob->gl_pathv); 87174918Speter pglob->gl_pathv = NULL; 8721573Srgrimes } 8731573Srgrimes} 8741573Srgrimes 8751573Srgrimesstatic DIR * 876159294Sdelphijg_opendir(Char *str, glob_t *pglob) 8771573Srgrimes{ 8781573Srgrimes char buf[MAXPATHLEN]; 8791573Srgrimes 8801573Srgrimes if (!*str) 8811573Srgrimes strcpy(buf, "."); 88274918Speter else { 88374921Speter if (g_Ctoc(str, buf, sizeof(buf))) 88474918Speter return (NULL); 88574918Speter } 8861573Srgrimes 8871573Srgrimes if (pglob->gl_flags & GLOB_ALTDIRFUNC) 888228755Seadler return ((*pglob->gl_opendir)(buf)); 8891573Srgrimes 890228755Seadler return (opendir(buf)); 8911573Srgrimes} 8921573Srgrimes 8931573Srgrimesstatic int 894159294Sdelphijg_lstat(Char *fn, struct stat *sb, glob_t *pglob) 8951573Srgrimes{ 8961573Srgrimes char buf[MAXPATHLEN]; 8971573Srgrimes 89874921Speter if (g_Ctoc(fn, buf, sizeof(buf))) { 89974918Speter errno = ENAMETOOLONG; 90074918Speter return (-1); 90174918Speter } 9021573Srgrimes if (pglob->gl_flags & GLOB_ALTDIRFUNC) 9031573Srgrimes return((*pglob->gl_lstat)(buf, sb)); 904228755Seadler return (lstat(buf, sb)); 9051573Srgrimes} 9061573Srgrimes 9071573Srgrimesstatic int 908159294Sdelphijg_stat(Char *fn, struct stat *sb, glob_t *pglob) 9091573Srgrimes{ 9101573Srgrimes char buf[MAXPATHLEN]; 9111573Srgrimes 91274921Speter if (g_Ctoc(fn, buf, sizeof(buf))) { 91374918Speter errno = ENAMETOOLONG; 91474918Speter return (-1); 91574918Speter } 9161573Srgrimes if (pglob->gl_flags & GLOB_ALTDIRFUNC) 917228755Seadler return ((*pglob->gl_stat)(buf, sb)); 918228755Seadler return (stat(buf, sb)); 9191573Srgrimes} 9201573Srgrimes 921180021Smtmstatic const Char * 922180021Smtmg_strchr(const Char *str, wchar_t ch) 9231573Srgrimes{ 924159294Sdelphij 9251573Srgrimes do { 9261573Srgrimes if (*str == ch) 9271573Srgrimes return (str); 9281573Srgrimes } while (*str++); 9291573Srgrimes return (NULL); 9301573Srgrimes} 9311573Srgrimes 93274918Speterstatic int 933159294Sdelphijg_Ctoc(const Char *str, char *buf, size_t len) 9341573Srgrimes{ 935132817Stjr mbstate_t mbs; 936132817Stjr size_t clen; 9371573Srgrimes 938132817Stjr memset(&mbs, 0, sizeof(mbs)); 939132817Stjr while (len >= MB_CUR_MAX) { 940132817Stjr clen = wcrtomb(buf, *str, &mbs); 941132817Stjr if (clen == (size_t)-1) 942132817Stjr return (1); 943132817Stjr if (*str == L'\0') 94474921Speter return (0); 945132817Stjr str++; 946132817Stjr buf += clen; 947132817Stjr len -= clen; 94874921Speter } 94974921Speter return (1); 9501573Srgrimes} 9511573Srgrimes 9521573Srgrimes#ifdef DEBUG 9538870Srgrimesstatic void 954159294Sdelphijqprintf(const char *str, Char *s) 9551573Srgrimes{ 95690045Sobrien Char *p; 9571573Srgrimes 9581573Srgrimes (void)printf("%s:\n", str); 9591573Srgrimes for (p = s; *p; p++) 9601573Srgrimes (void)printf("%c", CHAR(*p)); 9611573Srgrimes (void)printf("\n"); 9621573Srgrimes for (p = s; *p; p++) 9631573Srgrimes (void)printf("%c", *p & M_PROTECT ? '"' : ' '); 9641573Srgrimes (void)printf("\n"); 9651573Srgrimes for (p = s; *p; p++) 9661573Srgrimes (void)printf("%c", ismeta(*p) ? '_' : ' '); 9671573Srgrimes (void)printf("\n"); 9681573Srgrimes} 9691573Srgrimes#endif 970