lib/fnmatch.c

   1 /* Copyright (C) 1991,1992,1993,1996,1997,1998,1999,2000,2001,2002,2003,2004
   2         Free Software Foundation, Inc.
   3
   4    This program is free software; you can redistribute it and/or modify
   5    it under the terms of the GNU General Public License as published by
   6    the Free Software Foundation; either version 2, or (at your option)
   7    any later version.
   8
   9    This program is distributed in the hope that it will be useful,
  10    but WITHOUT ANY WARRANTY; without even the implied warranty of
  11    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12    GNU General Public License for more details.
  13
  14    You should have received a copy of the GNU General Public License
  15    along with this program; if not, write to the Free Software Foundation,
  16    Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
  17
  18 #if HAVE_CONFIG_H
  19 # include <config.h>
  20 #endif
  21
  22 /* Enable GNU extensions in fnmatch.h.  */
  23 #ifndef _GNU_SOURCE
  24 # define _GNU_SOURCE    1
  25 #endif
  26
  27 #if ! defined __builtin_expect && __GNUC__ < 3
  28 # define __builtin_expect(expr, expected) (expr)
  29 #endif
  30
  31 #include <fnmatch.h>
  32
  33 #include <alloca.h>
  34 #include <assert.h>
  35 #include <ctype.h>
  36 #include <errno.h>
  37 #include <stddef.h>
  38 #include <stdbool.h>
  39 #include <stdlib.h>
  40 #include <string.h>
  41
  42 #define WIDE_CHAR_SUPPORT \
  43   (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC \
  44    && HAVE_WMEMCHR && (HAVE_WMEMCPY || HAVE_WMEMPCPY))
  45
  46 /* For platform which support the ISO C amendement 1 functionality we
  47    support user defined character classes.  */
  48 #if defined _LIBC || WIDE_CHAR_SUPPORT
  49 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>.  */
  50 # include <wchar.h>
  51 # include <wctype.h>
  52 #endif
  53
  54 /* We need some of the locale data (the collation sequence information)
  55    but there is no interface to get this information in general.  Therefore
  56    we support a correct implementation only in glibc.  */
  57 #ifdef _LIBC
  58 # include "../locale/localeinfo.h"
  59 # include "../locale/elem-hash.h"
  60 # include "../locale/coll-lookup.h"
  61 # include <shlib-compat.h>
  62
  63 # define CONCAT(a,b) __CONCAT(a,b)
  64 # define mbsrtowcs __mbsrtowcs
  65 # define fnmatch __fnmatch
  66 extern int fnmatch (const char *pattern, const char *string, int flags);
  67 #endif
  68
  69 #ifndef SIZE_MAX
  70 # define SIZE_MAX ((size_t) -1)
  71 #endif
  72
  73 /* We often have to test for FNM_FILE_NAME and FNM_PERIOD being both set.  */
  74 #define NO_LEADING_PERIOD(flags) \
  75   ((flags & (FNM_FILE_NAME | FNM_PERIOD)) == (FNM_FILE_NAME | FNM_PERIOD))
  76
  77 /* Comment out all this code if we are using the GNU C Library, and are not
  78    actually compiling the library itself, and have not detected a bug
  79    in the library.  This code is part of the GNU C
  80    Library, but also included in many other GNU distributions.  Compiling
  81    and linking in this code is a waste when using the GNU C library
  82    (especially if it is a shared library).  Rather than having every GNU
  83    program understand `configure --with-gnu-libc' and omit the object files,
  84    it is simpler to just do this in the source for each such file.  */
  85
  86 #if defined _LIBC || !defined __GNU_LIBRARY__ || !HAVE_FNMATCH_GNU
  87
  88
  89 # if defined STDC_HEADERS || !defined isascii
  90 #  define ISASCII(c) 1
  91 # else
  92 #  define ISASCII(c) isascii(c)
  93 # endif
  94
  95 # ifdef isblank
  96 #  define ISBLANK(c) (ISASCII (c) && isblank (c))
  97 # else
  98 #  define ISBLANK(c) ((c) == ' ' || (c) == '\t')
  99 # endif
 100 # ifdef isgraph
 101 #  define ISGRAPH(c) (ISASCII (c) && isgraph (c))
 102 # else
 103 #  define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
 104 # endif
 105
 106 # define ISPRINT(c) (ISASCII (c) && isprint (c))
 107 # define ISDIGIT(c) (ISASCII (c) && isdigit (c))
 108 # define ISALNUM(c) (ISASCII (c) && isalnum (c))
 109 # define ISALPHA(c) (ISASCII (c) && isalpha (c))
 110 # define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
 111 # define ISLOWER(c) (ISASCII (c) && islower (c))
 112 # define ISPUNCT(c) (ISASCII (c) && ispunct (c))
 113 # define ISSPACE(c) (ISASCII (c) && isspace (c))
 114 # define ISUPPER(c) (ISASCII (c) && isupper (c))
 115 # define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
 116
 117 # define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
 118
 119 # if defined _LIBC || WIDE_CHAR_SUPPORT
 120 /* The GNU C library provides support for user-defined character classes
 121    and the functions from ISO C amendement 1.  */
 122 #  ifdef CHARCLASS_NAME_MAX
 123 #   define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
 124 #  else
 125 /* This shouldn't happen but some implementation might still have this
 126    problem.  Use a reasonable default value.  */
 127 #   define CHAR_CLASS_MAX_LENGTH 256
 128 #  endif
 129
 130 #  ifdef _LIBC
 131 #   define IS_CHAR_CLASS(string) __wctype (string)
 132 #  else
 133 #   define IS_CHAR_CLASS(string) wctype (string)
 134 #  endif
 135
 136 #  ifdef _LIBC
 137 #   define ISWCTYPE(WC, WT)     __iswctype (WC, WT)
 138 #  else
 139 #   define ISWCTYPE(WC, WT)     iswctype (WC, WT)
 140 #  endif
 141
 142 #  if (HAVE_MBSTATE_T && HAVE_MBSRTOWCS) || _LIBC
 143 /* In this case we are implementing the multibyte character handling.  */
 144 #   define HANDLE_MULTIBYTE     1
 145 #  endif
 146
 147 # else
 148 #  define CHAR_CLASS_MAX_LENGTH  6 /* Namely, `xdigit'.  */
 149
 150 #  define IS_CHAR_CLASS(string)                                               \
 151    (STREQ (string, "alpha") || STREQ (string, "upper")                        \
 152     || STREQ (string, "lower") || STREQ (string, "digit")                     \
 153     || STREQ (string, "alnum") || STREQ (string, "xdigit")                    \
 154     || STREQ (string, "space") || STREQ (string, "print")                     \
 155     || STREQ (string, "punct") || STREQ (string, "graph")                     \
 156     || STREQ (string, "cntrl") || STREQ (string, "blank"))
 157 # endif
 158
 159 /* Avoid depending on library functions or files
 160    whose names are inconsistent.  */
 161
 162 /* Global variable.  */
 163 static int posixly_correct;
 164
 165 # ifndef internal_function
 166 /* Inside GNU libc we mark some function in a special way.  In other
 167    environments simply ignore the marking.  */
 168 #  define internal_function
 169 # endif
 170
 171 /* Note that this evaluates C many times.  */
 172 # ifdef _LIBC
 173 #  define FOLD(c) ((flags & FNM_CASEFOLD) ? tolower (c) : (c))
 174 # else
 175 #  define FOLD(c) ((flags & FNM_CASEFOLD) && ISUPPER (c) ? tolower (c) : (c))
 176 # endif
 177 # define CHAR   char
 178 # define UCHAR  unsigned char
 179 # define INT    int
 180 # define FCT    internal_fnmatch
 181 # define EXT    ext_match
 182 # define END    end_pattern
 183 # define L(CS)  CS
 184 # ifdef _LIBC
 185 #  define BTOWC(C)      __btowc (C)
 186 # else
 187 #  define BTOWC(C)      btowc (C)
 188 # endif
 189 # define STRLEN(S) strlen (S)
 190 # define STRCAT(D, S) strcat (D, S)
 191 # ifdef _LIBC
 192 #  define MEMPCPY(D, S, N) __mempcpy (D, S, N)
 193 # else
 194 #  if HAVE_MEMPCPY
 195 #   define MEMPCPY(D, S, N) mempcpy (D, S, N)
 196 #  else
 197 #   define MEMPCPY(D, S, N) ((void *) ((char *) memcpy (D, S, N) + (N)))
 198 #  endif
 199 # endif
 200 # define MEMCHR(S, C, N) memchr (S, C, N)
 201 # define STRCOLL(S1, S2) strcoll (S1, S2)
 202 # include "fnmatch_loop.c"
 203
 204
 205 # if HANDLE_MULTIBYTE
 206 #  define FOLD(c) ((flags & FNM_CASEFOLD) ? towlower (c) : (c))
 207 #  define CHAR  wchar_t
 208 #  define UCHAR wint_t
 209 #  define INT   wint_t
 210 #  define FCT   internal_fnwmatch
 211 #  define EXT   ext_wmatch
 212 #  define END   end_wpattern
 213 #  define L(CS) L##CS
 214 #  define BTOWC(C)      (C)
 215 #  ifdef _LIBC
 216 #   define STRLEN(S) __wcslen (S)
 217 #   define STRCAT(D, S) __wcscat (D, S)
 218 #   define MEMPCPY(D, S, N) __wmempcpy (D, S, N)
 219 #  else
 220 #   define STRLEN(S) wcslen (S)
 221 #   define STRCAT(D, S) wcscat (D, S)
 222 #   if HAVE_WMEMPCPY
 223 #    define MEMPCPY(D, S, N) wmempcpy (D, S, N)
 224 #   else
 225 #    define MEMPCPY(D, S, N) (wmemcpy (D, S, N) + (N))
 226 #   endif
 227 #  endif
 228 #  define MEMCHR(S, C, N) wmemchr (S, C, N)
 229 #  define STRCOLL(S1, S2) wcscoll (S1, S2)
 230 #  define WIDE_CHAR_VERSION 1
 231
 232 #  undef IS_CHAR_CLASS
 233 /* We have to convert the wide character string in a multibyte string.  But
 234    we know that the character class names consist of alphanumeric characters
 235    from the portable character set, and since the wide character encoding
 236    for a member of the portable character set is the same code point as
 237    its single-byte encoding, we can use a simplified method to convert the
 238    string to a multibyte character string.  */
 239 static wctype_t
 240 is_char_class (const wchar_t *wcs)
 241 {
 242   char s[CHAR_CLASS_MAX_LENGTH + 1];
 243   char *cp = s;
 244
 245   do
 246     {
 247       /* Test for a printable character from the portable character set.  */
 248 #  ifdef _LIBC
 249       if (*wcs < 0x20 || *wcs > 0x7e
 250           || *wcs == 0x24 || *wcs == 0x40 || *wcs == 0x60)
 251         return (wctype_t) 0;
 252 #  else
 253       switch (*wcs)
 254         {
 255         case L' ': case L'!': case L'"': case L'#': case L'%':
 256         case L'&': case L'\'': case L'(': case L')': case L'*':
 257         case L'+': case L',': case L'-': case L'.': case L'/':
 258         case L'0': case L'1': case L'2': case L'3': case L'4':
 259         case L'5': case L'6': case L'7': case L'8': case L'9':
 260         case L':': case L';': case L'<': case L'=': case L'>':
 261         case L'?':
 262         case L'A': case L'B': case L'C': case L'D': case L'E':
 263         case L'F': case L'G': case L'H': case L'I': case L'J':
 264         case L'K': case L'L': case L'M': case L'N': case L'O':
 265         case L'P': case L'Q': case L'R': case L'S': case L'T':
 266         case L'U': case L'V': case L'W': case L'X': case L'Y':
 267         case L'Z':
 268         case L'[': case L'\\': case L']': case L'^': case L'_':
 269         case L'a': case L'b': case L'c': case L'd': case L'e':
 270         case L'f': case L'g': case L'h': case L'i': case L'j':
 271         case L'k': case L'l': case L'm': case L'n': case L'o':
 272         case L'p': case L'q': case L'r': case L's': case L't':
 273         case L'u': case L'v': case L'w': case L'x': case L'y':
 274         case L'z': case L'{': case L'|': case L'}': case L'~':
 275           break;
 276         default:
 277           return (wctype_t) 0;
 278         }
 279 #  endif
 280
 281       /* Avoid overrunning the buffer.  */
 282       if (cp == s + CHAR_CLASS_MAX_LENGTH)
 283         return (wctype_t) 0;
 284
 285       *cp++ = (char) *wcs++;
 286     }
 287   while (*wcs != L'\0');
 288
 289   *cp = '\0';
 290
 291 #  ifdef _LIBC
 292   return __wctype (s);
 293 #  else
 294   return wctype (s);
 295 #  endif
 296 }
 297 #  define IS_CHAR_CLASS(string) is_char_class (string)
 298
 299 #  include "fnmatch_loop.c"
 300 # endif
 301
 302
 303 int
 304 fnmatch (const char *pattern, const char *string, int flags)
 305 {
 306 # if HANDLE_MULTIBYTE
 307 #  define ALLOCA_LIMIT 2000
 308   if (__builtin_expect (MB_CUR_MAX, 1) != 1)
 309     {
 310       mbstate_t ps;
 311       size_t patsize;
 312       size_t strsize;
 313       size_t totsize;
 314       wchar_t *wpattern;
 315       wchar_t *wstring;
 316       int res;
 317
 318       /* Calculate the size needed to convert the strings to
 319          wide characters.  */
 320       memset (&ps, '\0', sizeof (ps));
 321       patsize = mbsrtowcs (NULL, &pattern, 0, &ps) + 1;
 322       if (__builtin_expect (patsize == 0, 0))
 323         /* Something wrong.
 324            XXX Do we have to set `errno' to something which mbsrtows hasn't
 325            already done?  */
 326         return -1;
 327       assert (mbsinit (&ps));
 328       strsize = mbsrtowcs (NULL, &string, 0, &ps) + 1;
 329       if (__builtin_expect (strsize == 0, 0))
 330         /* Something wrong.
 331            XXX Do we have to set `errno' to something which mbsrtows hasn't
 332            already done?  */
 333         return -1;
 334       assert (mbsinit (&ps));
 335       totsize = patsize + strsize;
 336       if (__builtin_expect (! (patsize <= totsize
 337                                && totsize <= SIZE_MAX / sizeof (wchar_t)),
 338                             0))
 339         {
 340           errno = ENOMEM;
 341           return -1;
 342         }
 343
 344       /* Allocate room for the wide characters.  */
 345       if (__builtin_expect (totsize < ALLOCA_LIMIT, 1))
 346         wpattern = (wchar_t *) alloca (totsize * sizeof (wchar_t));
 347       else
 348         {
 349           wpattern = malloc (totsize * sizeof (wchar_t));
 350           if (__builtin_expect (! wpattern, 0))
 351             {
 352               errno = ENOMEM;
 353               return -1;
 354             }
 355         }
 356       wstring = wpattern + patsize;
 357
 358       /* Convert the strings into wide characters.  */
 359       mbsrtowcs (wpattern, &pattern, patsize, &ps);
 360       assert (mbsinit (&ps));
 361       mbsrtowcs (wstring, &string, strsize, &ps);
 362
 363       res = internal_fnwmatch (wpattern, wstring, wstring + strsize - 1,
 364                                flags & FNM_PERIOD, flags);
 365
 366       if (__builtin_expect (! (totsize < ALLOCA_LIMIT), 0))
 367         free (wpattern);
 368       return res;
 369     }
 370 # endif /* HANDLE_MULTIBYTE */
 371
 372   return internal_fnmatch (pattern, string, string + strlen (string),
 373                            flags & FNM_PERIOD, flags);
 374 }
 375
 376 # ifdef _LIBC
 377 #  undef fnmatch
 378 versioned_symbol (libc, __fnmatch, fnmatch, GLIBC_2_2_3);
 379 #  if SHLIB_COMPAT(libc, GLIBC_2_0, GLIBC_2_2_3)
 380 strong_alias (__fnmatch, __fnmatch_old)
 381 compat_symbol (libc, __fnmatch_old, fnmatch, GLIBC_2_0);
 382 #  endif
 383 libc_hidden_ver (__fnmatch, fnmatch)
 384 # endif
 385
 386 #endif  /* _LIBC or not __GNU_LIBRARY__.  */