This is the beginning of series of patches with addition
[glibc.git] / posix / fnmatch.c
bloba707847dccd5945a1266bff6942b18f2fb9144bf
1 /* Copyright (C) 1991-2015 Free Software Foundation, Inc.
2 This file is part of the GNU C Library.
4 The GNU C Library is free software; you can redistribute it and/or
5 modify it under the terms of the GNU Lesser General Public
6 License as published by the Free Software Foundation; either
7 version 2.1 of the License, or (at your option) any later version.
9 The GNU C Library 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 GNU
12 Lesser General Public License for more details.
14 You should have received a copy of the GNU Lesser General Public
15 License along with the GNU C Library; if not, see
16 <http://www.gnu.org/licenses/>. */
18 #if HAVE_CONFIG_H
19 # include <config.h>
20 #endif
22 /* Enable GNU extensions in fnmatch.h. */
23 #ifndef _GNU_SOURCE
24 # define _GNU_SOURCE 1
25 #endif
27 #include <assert.h>
28 #include <errno.h>
29 #include <fnmatch.h>
30 #include <ctype.h>
31 #include <string.h>
33 #if defined STDC_HEADERS || defined _LIBC
34 # include <stdlib.h>
35 #endif
37 #ifdef _LIBC
38 # include <alloca.h>
39 #else
40 # define alloca_account(size., var) alloca (size)
41 #endif
43 /* For platform which support the ISO C amendement 1 functionality we
44 support user defined character classes. */
45 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
46 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
47 # include <wchar.h>
48 # include <wctype.h>
49 #endif
51 /* We need some of the locale data (the collation sequence information)
52 but there is no interface to get this information in general. Therefore
53 we support a correct implementation only in glibc. */
54 #ifdef _LIBC
55 # include "../locale/localeinfo.h"
56 # include "../locale/elem-hash.h"
57 # include "../locale/coll-lookup.h"
58 # include <shlib-compat.h>
60 # define CONCAT(a,b) __CONCAT(a,b)
61 # define mbsrtowcs __mbsrtowcs
62 # define fnmatch __fnmatch
63 extern int fnmatch (const char *pattern, const char *string, int flags);
64 #endif
66 /* We often have to test for FNM_FILE_NAME and FNM_PERIOD being both set. */
67 #define NO_LEADING_PERIOD(flags) \
68 ((flags & (FNM_FILE_NAME | FNM_PERIOD)) == (FNM_FILE_NAME | FNM_PERIOD))
70 /* Comment out all this code if we are using the GNU C Library, and are not
71 actually compiling the library itself. This code is part of the GNU C
72 Library, but also included in many other GNU distributions. Compiling
73 and linking in this code is a waste when using the GNU C library
74 (especially if it is a shared library). Rather than having every GNU
75 program understand `configure --with-gnu-libc' and omit the object files,
76 it is simpler to just do this in the source for each such file. */
78 #if defined _LIBC || !defined __GNU_LIBRARY__
81 # if defined STDC_HEADERS || !defined isascii
82 # define ISASCII(c) 1
83 # else
84 # define ISASCII(c) isascii(c)
85 # endif
87 # ifdef isblank
88 # define ISBLANK(c) (ISASCII (c) && isblank (c))
89 # else
90 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
91 # endif
92 # ifdef isgraph
93 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
94 # else
95 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
96 # endif
98 # define ISPRINT(c) (ISASCII (c) && isprint (c))
99 # define ISDIGIT(c) (ISASCII (c) && isdigit (c))
100 # define ISALNUM(c) (ISASCII (c) && isalnum (c))
101 # define ISALPHA(c) (ISASCII (c) && isalpha (c))
102 # define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
103 # define ISLOWER(c) (ISASCII (c) && islower (c))
104 # define ISPUNCT(c) (ISASCII (c) && ispunct (c))
105 # define ISSPACE(c) (ISASCII (c) && isspace (c))
106 # define ISUPPER(c) (ISASCII (c) && isupper (c))
107 # define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
109 # define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
111 # if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
112 /* The GNU C library provides support for user-defined character classes
113 and the functions from ISO C amendement 1. */
114 # ifdef CHARCLASS_NAME_MAX
115 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
116 # else
117 /* This shouldn't happen but some implementation might still have this
118 problem. Use a reasonable default value. */
119 # define CHAR_CLASS_MAX_LENGTH 256
120 # endif
122 # ifdef _LIBC
123 # define IS_CHAR_CLASS(string) __wctype (string)
124 # else
125 # define IS_CHAR_CLASS(string) wctype (string)
126 # endif
128 # ifdef _LIBC
129 # define ISWCTYPE(WC, WT) __iswctype (WC, WT)
130 # else
131 # define ISWCTYPE(WC, WT) iswctype (WC, WT)
132 # endif
134 # if (HAVE_MBSTATE_T && HAVE_MBSRTOWCS) || _LIBC
135 /* In this case we are implementing the multibyte character handling. */
136 # define HANDLE_MULTIBYTE 1
137 # endif
139 # else
140 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
142 # define IS_CHAR_CLASS(string) \
143 (STREQ (string, "alpha") || STREQ (string, "upper") \
144 || STREQ (string, "lower") || STREQ (string, "digit") \
145 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
146 || STREQ (string, "space") || STREQ (string, "print") \
147 || STREQ (string, "punct") || STREQ (string, "graph") \
148 || STREQ (string, "cntrl") || STREQ (string, "blank"))
149 # endif
151 /* Avoid depending on library functions or files
152 whose names are inconsistent. */
154 # if !defined _LIBC && !defined getenv
155 extern char *getenv ();
156 # endif
158 # ifndef errno
159 extern int errno;
160 # endif
162 /* Global variable. */
163 static int posixly_correct;
165 /* This function doesn't exist on most systems. */
167 # if !defined HAVE___STRCHRNUL && !defined _LIBC
168 static char *
169 __strchrnul (s, c)
170 const char *s;
171 int c;
173 char *result = strchr (s, c);
174 if (result == NULL)
175 result = strchr (s, '\0');
176 return result;
178 # endif
180 # if HANDLE_MULTIBYTE && !defined HAVE___STRCHRNUL && !defined _LIBC
181 static wchar_t *
182 __wcschrnul (s, c)
183 const wchar_t *s;
184 wint_t c;
186 wchar_t *result = wcschr (s, c);
187 if (result == NULL)
188 result = wcschr (s, '\0');
189 return result;
191 # endif
193 # ifndef internal_function
194 /* Inside GNU libc we mark some function in a special way. In other
195 environments simply ignore the marking. */
196 # define internal_function
197 # endif
199 /* Note that this evaluates C many times. */
200 # ifdef _LIBC
201 # define FOLD(c) ((flags & FNM_CASEFOLD) ? tolower (c) : (c))
202 # else
203 # define FOLD(c) ((flags & FNM_CASEFOLD) && ISUPPER (c) ? tolower (c) : (c))
204 # endif
205 # define CHAR char
206 # define UCHAR unsigned char
207 # define INT int
208 # define FCT internal_fnmatch
209 # define EXT ext_match
210 # define END end_pattern
211 # define STRUCT fnmatch_struct
212 # define L(CS) CS
213 # ifdef _LIBC
214 # define BTOWC(C) __btowc (C)
215 # else
216 # define BTOWC(C) btowc (C)
217 # endif
218 # define STRLEN(S) strlen (S)
219 # define STRCAT(D, S) strcat (D, S)
220 # define MEMPCPY(D, S, N) __mempcpy (D, S, N)
221 # define MEMCHR(S, C, N) memchr (S, C, N)
222 # define STRCOLL(S1, S2) strcoll (S1, S2)
223 # define WIDE_CHAR_VERSION 0
224 # include <locale/weight.h>
225 # define FINDIDX findidx
226 # include "fnmatch_loop.c"
229 # if HANDLE_MULTIBYTE
230 /* Note that this evaluates C many times. */
231 # ifdef _LIBC
232 # define FOLD(c) ((flags & FNM_CASEFOLD) ? towlower (c) : (c))
233 # else
234 # define FOLD(c) ((flags & FNM_CASEFOLD) && ISUPPER (c) ? towlower (c) : (c))
235 # endif
236 # define CHAR wchar_t
237 # define UCHAR wint_t
238 # define INT wint_t
239 # define FCT internal_fnwmatch
240 # define EXT ext_wmatch
241 # define END end_wpattern
242 # define STRUCT fnwmatch_struct
243 # define L(CS) L##CS
244 # define BTOWC(C) (C)
245 # define STRLEN(S) __wcslen (S)
246 # define STRCAT(D, S) __wcscat (D, S)
247 # define MEMPCPY(D, S, N) __wmempcpy (D, S, N)
248 # define MEMCHR(S, C, N) wmemchr (S, C, N)
249 # define STRCOLL(S1, S2) wcscoll (S1, S2)
250 # define WIDE_CHAR_VERSION 1
251 /* Change the name the header defines so it doesn't conflict with
252 the <locale/weight.h> version included above. */
253 # define findidx findidxwc
254 # include <locale/weightwc.h>
255 # undef findidx
256 # define FINDIDX findidxwc
258 # undef IS_CHAR_CLASS
259 /* We have to convert the wide character string in a multibyte string. But
260 we know that the character class names consist of alphanumeric characters
261 from the portable character set, and since the wide character encoding
262 for a member of the portable character set is the same code point as
263 its single-byte encoding, we can use a simplified method to convert the
264 string to a multibyte character string. */
265 static wctype_t
266 is_char_class (const wchar_t *wcs)
268 char s[CHAR_CLASS_MAX_LENGTH + 1];
269 char *cp = s;
273 /* Test for a printable character from the portable character set. */
274 # ifdef _LIBC
275 if (*wcs < 0x20 || *wcs > 0x7e
276 || *wcs == 0x24 || *wcs == 0x40 || *wcs == 0x60)
277 return (wctype_t) 0;
278 # else
279 switch (*wcs)
281 case L' ': case L'!': case L'"': case L'#': case L'%':
282 case L'&': case L'\'': case L'(': case L')': case L'*':
283 case L'+': case L',': case L'-': case L'.': case L'/':
284 case L'0': case L'1': case L'2': case L'3': case L'4':
285 case L'5': case L'6': case L'7': case L'8': case L'9':
286 case L':': case L';': case L'<': case L'=': case L'>':
287 case L'?':
288 case L'A': case L'B': case L'C': case L'D': case L'E':
289 case L'F': case L'G': case L'H': case L'I': case L'J':
290 case L'K': case L'L': case L'M': case L'N': case L'O':
291 case L'P': case L'Q': case L'R': case L'S': case L'T':
292 case L'U': case L'V': case L'W': case L'X': case L'Y':
293 case L'Z':
294 case L'[': case L'\\': case L']': case L'^': case L'_':
295 case L'a': case L'b': case L'c': case L'd': case L'e':
296 case L'f': case L'g': case L'h': case L'i': case L'j':
297 case L'k': case L'l': case L'm': case L'n': case L'o':
298 case L'p': case L'q': case L'r': case L's': case L't':
299 case L'u': case L'v': case L'w': case L'x': case L'y':
300 case L'z': case L'{': case L'|': case L'}': case L'~':
301 break;
302 default:
303 return (wctype_t) 0;
305 # endif
307 /* Avoid overrunning the buffer. */
308 if (cp == s + CHAR_CLASS_MAX_LENGTH)
309 return (wctype_t) 0;
311 *cp++ = (char) *wcs++;
313 while (*wcs != L'\0');
315 *cp = '\0';
317 # ifdef _LIBC
318 return __wctype (s);
319 # else
320 return wctype (s);
321 # endif
323 # define IS_CHAR_CLASS(string) is_char_class (string)
325 # include "fnmatch_loop.c"
326 # endif
330 fnmatch (pattern, string, flags)
331 const char *pattern;
332 const char *string;
333 int flags;
335 # if HANDLE_MULTIBYTE
336 if (__builtin_expect (MB_CUR_MAX, 1) != 1)
338 mbstate_t ps;
339 size_t n;
340 const char *p;
341 wchar_t *wpattern_malloc = NULL;
342 wchar_t *wpattern;
343 wchar_t *wstring_malloc = NULL;
344 wchar_t *wstring;
345 size_t alloca_used = 0;
347 /* Convert the strings into wide characters. */
348 memset (&ps, '\0', sizeof (ps));
349 p = pattern;
350 #ifdef _LIBC
351 n = strnlen (pattern, 1024);
352 #else
353 n = strlen (pattern);
354 #endif
355 if (__glibc_likely (n < 1024))
357 wpattern = (wchar_t *) alloca_account ((n + 1) * sizeof (wchar_t),
358 alloca_used);
359 n = mbsrtowcs (wpattern, &p, n + 1, &ps);
360 if (__glibc_unlikely (n == (size_t) -1))
361 /* Something wrong.
362 XXX Do we have to set `errno' to something which mbsrtows hasn't
363 already done? */
364 return -1;
365 if (p)
367 memset (&ps, '\0', sizeof (ps));
368 goto prepare_wpattern;
371 else
373 prepare_wpattern:
374 n = mbsrtowcs (NULL, &pattern, 0, &ps);
375 if (__glibc_unlikely (n == (size_t) -1))
376 /* Something wrong.
377 XXX Do we have to set `errno' to something which mbsrtows hasn't
378 already done? */
379 return -1;
380 if (__glibc_unlikely (n >= (size_t) -1 / sizeof (wchar_t)))
382 __set_errno (ENOMEM);
383 return -2;
385 wpattern_malloc = wpattern
386 = (wchar_t *) malloc ((n + 1) * sizeof (wchar_t));
387 assert (mbsinit (&ps));
388 if (wpattern == NULL)
389 return -2;
390 (void) mbsrtowcs (wpattern, &pattern, n + 1, &ps);
393 assert (mbsinit (&ps));
394 #ifdef _LIBC
395 n = strnlen (string, 1024);
396 #else
397 n = strlen (string);
398 #endif
399 p = string;
400 if (__glibc_likely (n < 1024))
402 wstring = (wchar_t *) alloca_account ((n + 1) * sizeof (wchar_t),
403 alloca_used);
404 n = mbsrtowcs (wstring, &p, n + 1, &ps);
405 if (__glibc_unlikely (n == (size_t) -1))
407 /* Something wrong.
408 XXX Do we have to set `errno' to something which
409 mbsrtows hasn't already done? */
410 free_return:
411 free (wpattern_malloc);
412 return -1;
414 if (p)
416 memset (&ps, '\0', sizeof (ps));
417 goto prepare_wstring;
420 else
422 prepare_wstring:
423 n = mbsrtowcs (NULL, &string, 0, &ps);
424 if (__glibc_unlikely (n == (size_t) -1))
425 /* Something wrong.
426 XXX Do we have to set `errno' to something which mbsrtows hasn't
427 already done? */
428 goto free_return;
429 if (__glibc_unlikely (n >= (size_t) -1 / sizeof (wchar_t)))
431 free (wpattern_malloc);
432 __set_errno (ENOMEM);
433 return -2;
436 wstring_malloc = wstring
437 = (wchar_t *) malloc ((n + 1) * sizeof (wchar_t));
438 if (wstring == NULL)
440 free (wpattern_malloc);
441 return -2;
443 assert (mbsinit (&ps));
444 (void) mbsrtowcs (wstring, &string, n + 1, &ps);
447 int res = internal_fnwmatch (wpattern, wstring, wstring + n,
448 flags & FNM_PERIOD, flags, NULL,
449 alloca_used);
451 free (wstring_malloc);
452 free (wpattern_malloc);
454 return res;
456 # endif /* mbstate_t and mbsrtowcs or _LIBC. */
458 return internal_fnmatch (pattern, string, string + strlen (string),
459 flags & FNM_PERIOD, flags, NULL, 0);
462 # ifdef _LIBC
463 # undef fnmatch
464 versioned_symbol (libc, __fnmatch, fnmatch, GLIBC_2_2_3);
465 # if SHLIB_COMPAT(libc, GLIBC_2_0, GLIBC_2_2_3)
466 strong_alias (__fnmatch, __fnmatch_old)
467 compat_symbol (libc, __fnmatch_old, fnmatch, GLIBC_2_0);
468 # endif
469 libc_hidden_ver (__fnmatch, fnmatch)
470 # endif
472 #endif /* _LIBC or not __GNU_LIBRARY__. */