1 /* Copyright (C) 1991,1992,1993,1996,1997,1998,1999,2000,2001,2002,2003,2004,2005,2006
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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
17
18 #ifndef _LIBC
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 isblank || HAVE_DECL_ISBLANK)
90 # define isblank(c) ((c) == ' ' || (c) == '\t')
91 # endif
92
93 # define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
94
95 # if defined _LIBC || WIDE_CHAR_SUPPORT
96 /* The GNU C library provides support for user-defined character classes
97 and the functions from ISO C amendement 1. */
98 # ifdef CHARCLASS_NAME_MAX
99 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
100 # else
101 /* This shouldn't happen but some implementation might still have this
102 problem. Use a reasonable default value. */
103 # define CHAR_CLASS_MAX_LENGTH 256
104 # endif
105
106 # ifdef _LIBC
107 # define IS_CHAR_CLASS(string) __wctype (string)
108 # else
109 # define IS_CHAR_CLASS(string) wctype (string)
110 # endif
111
112 # ifdef _LIBC
113 # define ISWCTYPE(WC, WT) __iswctype (WC, WT)
114 # else
115 # define ISWCTYPE(WC, WT) iswctype (WC, WT)
116 # endif
117
118 # if (HAVE_MBSTATE_T && HAVE_MBSRTOWCS) || _LIBC
119 /* In this case we are implementing the multibyte character handling. */
120 # define HANDLE_MULTIBYTE 1
121 # endif
122
123 # else
124 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
125
126 # define IS_CHAR_CLASS(string) \
127 (STREQ (string, "alpha") || STREQ (string, "upper") \
128 || STREQ (string, "lower") || STREQ (string, "digit") \
129 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
130 || STREQ (string, "space") || STREQ (string, "print") \
131 || STREQ (string, "punct") || STREQ (string, "graph") \
132 || STREQ (string, "cntrl") || STREQ (string, "blank"))
133 # endif
134
135 /* Avoid depending on library functions or files
136 whose names are inconsistent. */
137
138 /* Global variable. */
139 static int posixly_correct;
140
141 # ifndef internal_function
142 /* Inside GNU libc we mark some function in a special way. In other
143 environments simply ignore the marking. */
144 # define internal_function
145 # endif
146
147 /* Note that this evaluates C many times. */
148 # define FOLD(c) ((flags & FNM_CASEFOLD) ? tolower (c) : (c))
149 # define CHAR char
150 # define UCHAR unsigned char
151 # define INT int
152 # define FCT internal_fnmatch
153 # define EXT ext_match
154 # define END end_pattern
155 # define L_(CS) CS
156 # ifdef _LIBC
157 # define BTOWC(C) __btowc (C)
158 # else
159 # define BTOWC(C) btowc (C)
160 # endif
161 # define STRLEN(S) strlen (S)
162 # define STRCAT(D, S) strcat (D, S)
163 # ifdef _LIBC
164 # define MEMPCPY(D, S, N) __mempcpy (D, S, N)
165 # else
166 # if HAVE_MEMPCPY
167 # define MEMPCPY(D, S, N) mempcpy (D, S, N)
168 # else
169 # define MEMPCPY(D, S, N) ((void *) ((char *) memcpy (D, S, N) + (N)))
170 # endif
171 # endif
172 # define MEMCHR(S, C, N) memchr (S, C, N)
173 # define STRCOLL(S1, S2) strcoll (S1, S2)
174 # include "fnmatch_loop.c"
175
176
177 # if HANDLE_MULTIBYTE
178 # define FOLD(c) ((flags & FNM_CASEFOLD) ? towlower (c) : (c))
179 # define CHAR wchar_t
180 # define UCHAR wint_t
181 # define INT wint_t
182 # define FCT internal_fnwmatch
183 # define EXT ext_wmatch
184 # define END end_wpattern
185 # define L_(CS) L##CS
186 # define BTOWC(C) (C)
187 # ifdef _LIBC
188 # define STRLEN(S) __wcslen (S)
189 # define STRCAT(D, S) __wcscat (D, S)
190 # define MEMPCPY(D, S, N) __wmempcpy (D, S, N)
191 # else
192 # define STRLEN(S) wcslen (S)
193 # define STRCAT(D, S) wcscat (D, S)
194 # if HAVE_WMEMPCPY
195 # define MEMPCPY(D, S, N) wmempcpy (D, S, N)
196 # else
197 # define MEMPCPY(D, S, N) (wmemcpy (D, S, N) + (N))
198 # endif
199 # endif
200 # define MEMCHR(S, C, N) wmemchr (S, C, N)
201 # define STRCOLL(S1, S2) wcscoll (S1, S2)
202 # define WIDE_CHAR_VERSION 1
203
204 # undef IS_CHAR_CLASS
205 /* We have to convert the wide character string in a multibyte string. But
206 we know that the character class names consist of alphanumeric characters
207 from the portable character set, and since the wide character encoding
208 for a member of the portable character set is the same code point as
209 its single-byte encoding, we can use a simplified method to convert the
210 string to a multibyte character string. */
211 static wctype_t
is_char_class(const wchar_t * wcs)212 is_char_class (const wchar_t *wcs)
213 {
214 char s[CHAR_CLASS_MAX_LENGTH + 1];
215 char *cp = s;
216
217 do
218 {
219 /* Test for a printable character from the portable character set. */
220 # ifdef _LIBC
221 if (*wcs < 0x20 || *wcs > 0x7e
222 || *wcs == 0x24 || *wcs == 0x40 || *wcs == 0x60)
223 return (wctype_t) 0;
224 # else
225 switch (*wcs)
226 {
227 case L' ': case L'!': case L'"': case L'#': case L'%':
228 case L'&': case L'\'': case L'(': case L')': case L'*':
229 case L'+': case L',': case L'-': case L'.': case L'/':
230 case L'0': case L'1': case L'2': case L'3': case L'4':
231 case L'5': case L'6': case L'7': case L'8': case L'9':
232 case L':': case L';': case L'<': case L'=': case L'>':
233 case L'?':
234 case L'A': case L'B': case L'C': case L'D': case L'E':
235 case L'F': case L'G': case L'H': case L'I': case L'J':
236 case L'K': case L'L': case L'M': case L'N': case L'O':
237 case L'P': case L'Q': case L'R': case L'S': case L'T':
238 case L'U': case L'V': case L'W': case L'X': case L'Y':
239 case L'Z':
240 case L'[': case L'\\': case L']': case L'^': case L'_':
241 case L'a': case L'b': case L'c': case L'd': case L'e':
242 case L'f': case L'g': case L'h': case L'i': case L'j':
243 case L'k': case L'l': case L'm': case L'n': case L'o':
244 case L'p': case L'q': case L'r': case L's': case L't':
245 case L'u': case L'v': case L'w': case L'x': case L'y':
246 case L'z': case L'{': case L'|': case L'}': case L'~':
247 break;
248 default:
249 return (wctype_t) 0;
250 }
251 # endif
252
253 /* Avoid overrunning the buffer. */
254 if (cp == s + CHAR_CLASS_MAX_LENGTH)
255 return (wctype_t) 0;
256
257 *cp++ = (char) *wcs++;
258 }
259 while (*wcs != L'\0');
260
261 *cp = '\0';
262
263 # ifdef _LIBC
264 return __wctype (s);
265 # else
266 return wctype (s);
267 # endif
268 }
269 # define IS_CHAR_CLASS(string) is_char_class (string)
270
271 # include "fnmatch_loop.c"
272 # endif
273
274
275 int
fnmatch(const char * pattern,const char * string,int flags)276 fnmatch (const char *pattern, const char *string, int flags)
277 {
278 # if HANDLE_MULTIBYTE
279 # if HAVE_ALLOCA || defined _LIBC
280 # define ALLOCA_LIMIT 2000
281 # else
282 # define ALLOCA_LIMIT 0
283 # endif
284 if (__builtin_expect (MB_CUR_MAX, 1) != 1)
285 {
286 mbstate_t ps;
287 size_t patsize;
288 size_t strsize;
289 size_t totsize;
290 wchar_t *wpattern;
291 wchar_t *wstring;
292 int res;
293
294 /* Calculate the size needed to convert the strings to
295 wide characters. */
296 memset (&ps, '\0', sizeof (ps));
297 patsize = mbsrtowcs (NULL, &pattern, 0, &ps) + 1;
298 if (__builtin_expect (patsize != 0, 1))
299 {
300 assert (mbsinit (&ps));
301 strsize = mbsrtowcs (NULL, &string, 0, &ps) + 1;
302 if (__builtin_expect (strsize != 0, 1))
303 {
304 assert (mbsinit (&ps));
305 totsize = patsize + strsize;
306 if (__builtin_expect (! (patsize <= totsize
307 && totsize <= SIZE_MAX / sizeof (wchar_t)),
308 0))
309 {
310 errno = ENOMEM;
311 return -1;
312 }
313
314 /* Allocate room for the wide characters. */
315 if (__builtin_expect (totsize < ALLOCA_LIMIT, 1))
316 wpattern = (wchar_t *) alloca (totsize * sizeof (wchar_t));
317 else
318 {
319 wpattern = malloc (totsize * sizeof (wchar_t));
320 if (__builtin_expect (! wpattern, 0))
321 {
322 errno = ENOMEM;
323 return -1;
324 }
325 }
326 wstring = wpattern + patsize;
327
328 /* Convert the strings into wide characters. */
329 mbsrtowcs (wpattern, &pattern, patsize, &ps);
330 assert (mbsinit (&ps));
331 mbsrtowcs (wstring, &string, strsize, &ps);
332
333 res = internal_fnwmatch (wpattern, wstring, wstring + strsize - 1,
334 flags & FNM_PERIOD, flags);
335
336 if (__builtin_expect (! (totsize < ALLOCA_LIMIT), 0))
337 free (wpattern);
338 return res;
339 }
340 }
341 }
342
343 # endif /* HANDLE_MULTIBYTE */
344
345 return internal_fnmatch (pattern, string, string + strlen (string),
346 flags & FNM_PERIOD, flags);
347 }
348
349 # ifdef _LIBC
350 # undef fnmatch
351 versioned_symbol (libc, __fnmatch, fnmatch, GLIBC_2_2_3);
352 # if SHLIB_COMPAT(libc, GLIBC_2_0, GLIBC_2_2_3)
353 strong_alias (__fnmatch, __fnmatch_old)
354 compat_symbol (libc, __fnmatch_old, fnmatch, GLIBC_2_0);
355 # endif
356 libc_hidden_ver (__fnmatch, fnmatch)
357 # endif
358
359 #endif /* _LIBC or not __GNU_LIBRARY__. */
360