1/* Copyright (C) 1991-2017 Free Software Foundation, Inc.
2 This file is part of the GNU C Library.
3
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.
8
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.
13
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/>. */
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#include <assert.h>
28#include <errno.h>
29#include <fnmatch.h>
30#include <ctype.h>
31#include <string.h>
32
33#if defined STDC_HEADERS || defined _LIBC
34# include <stdlib.h>
35#endif
36
37#ifdef _LIBC
38# include <alloca.h>
39#else
40# define alloca_account(size., var) alloca (size)
41#endif
42
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
50
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>
59
60# define CONCAT(a,b) __CONCAT(a,b)
61# define mbsrtowcs __mbsrtowcs
62# define fnmatch __fnmatch
63extern int fnmatch (const char *pattern, const char *string, int flags);
64#endif
65
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))
69
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. */
77
78#if defined _LIBC || !defined __GNU_LIBRARY__
79
80
81# if defined STDC_HEADERS || !defined isascii
82# define ISASCII(c) 1
83# else
84# define ISASCII(c) isascii(c)
85# endif
86
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
97
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))
108
109# define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
110
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
121
122# ifdef _LIBC
123# define IS_CHAR_CLASS(string) __wctype (string)
124# else
125# define IS_CHAR_CLASS(string) wctype (string)
126# endif
127
128# ifdef _LIBC
129# define ISWCTYPE(WC, WT) __iswctype (WC, WT)
130# else
131# define ISWCTYPE(WC, WT) iswctype (WC, WT)
132# endif
133
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
138
139# else
140# define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
141
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
150
151/* Avoid depending on library functions or files
152 whose names are inconsistent. */
153
154# if !defined _LIBC && !defined getenv
155extern char *getenv ();
156# endif
157
158# ifndef errno
159extern int errno;
160# endif
161
162/* Global variable. */
163static int posixly_correct;
164
165/* This function doesn't exist on most systems. */
166
167# if !defined HAVE___STRCHRNUL && !defined _LIBC
168static char *
169__strchrnul (const char *s, int c)
170{
171 char *result = strchr (s, c);
172 if (result == NULL)
173 result = strchr (s, '\0');
174 return result;
175}
176# endif
177
178# if HANDLE_MULTIBYTE && !defined HAVE___STRCHRNUL && !defined _LIBC
179static wchar_t *
180__wcschrnul (const wchar_t *s, wint_t c)
181{
182 wchar_t *result = wcschr (s, c);
183 if (result == NULL)
184 result = wcschr (s, '\0');
185 return result;
186}
187# endif
188
189# ifndef internal_function
190/* Inside GNU libc we mark some function in a special way. In other
191 environments simply ignore the marking. */
192# define internal_function
193# endif
194
195/* Note that this evaluates C many times. */
196# ifdef _LIBC
197# define FOLD(c) ((flags & FNM_CASEFOLD) ? tolower (c) : (c))
198# else
199# define FOLD(c) ((flags & FNM_CASEFOLD) && ISUPPER (c) ? tolower (c) : (c))
200# endif
201# define CHAR char
202# define UCHAR unsigned char
203# define INT int
204# define FCT internal_fnmatch
205# define EXT ext_match
206# define END end_pattern
207# define STRUCT fnmatch_struct
208# define L(CS) CS
209# ifdef _LIBC
210# define BTOWC(C) __btowc (C)
211# else
212# define BTOWC(C) btowc (C)
213# endif
214# define STRLEN(S) strlen (S)
215# define STRCAT(D, S) strcat (D, S)
216# define MEMPCPY(D, S, N) __mempcpy (D, S, N)
217# define MEMCHR(S, C, N) memchr (S, C, N)
218# define STRCOLL(S1, S2) strcoll (S1, S2)
219# define WIDE_CHAR_VERSION 0
220# include <locale/weight.h>
221# define FINDIDX findidx
222# include "fnmatch_loop.c"
223
224
225# if HANDLE_MULTIBYTE
226/* Note that this evaluates C many times. */
227# ifdef _LIBC
228# define FOLD(c) ((flags & FNM_CASEFOLD) ? __towlower (c) : (c))
229# else
230# define FOLD(c) ((flags & FNM_CASEFOLD) && ISUPPER (c) ? towlower (c) : (c))
231# endif
232# define CHAR wchar_t
233# define UCHAR wint_t
234# define INT wint_t
235# define FCT internal_fnwmatch
236# define EXT ext_wmatch
237# define END end_wpattern
238# define STRUCT fnwmatch_struct
239# define L(CS) L##CS
240# define BTOWC(C) (C)
241# define STRLEN(S) __wcslen (S)
242# define STRCAT(D, S) __wcscat (D, S)
243# define MEMPCPY(D, S, N) __wmempcpy (D, S, N)
244# define MEMCHR(S, C, N) __wmemchr (S, C, N)
245# define STRCOLL(S1, S2) wcscoll (S1, S2)
246# define WIDE_CHAR_VERSION 1
247/* Change the name the header defines so it doesn't conflict with
248 the <locale/weight.h> version included above. */
249# define findidx findidxwc
250# include <locale/weightwc.h>
251# undef findidx
252# define FINDIDX findidxwc
253
254# undef IS_CHAR_CLASS
255/* We have to convert the wide character string in a multibyte string. But
256 we know that the character class names consist of alphanumeric characters
257 from the portable character set, and since the wide character encoding
258 for a member of the portable character set is the same code point as
259 its single-byte encoding, we can use a simplified method to convert the
260 string to a multibyte character string. */
261static wctype_t
262is_char_class (const wchar_t *wcs)
263{
264 char s[CHAR_CLASS_MAX_LENGTH + 1];
265 char *cp = s;
266
267 do
268 {
269 /* Test for a printable character from the portable character set. */
270# ifdef _LIBC
271 if (*wcs < 0x20 || *wcs > 0x7e
272 || *wcs == 0x24 || *wcs == 0x40 || *wcs == 0x60)
273 return (wctype_t) 0;
274# else
275 switch (*wcs)
276 {
277 case L' ': case L'!': case L'"': case L'#': case L'%':
278 case L'&': case L'\'': case L'(': case L')': case L'*':
279 case L'+': case L',': case L'-': case L'.': case L'/':
280 case L'0': case L'1': case L'2': case L'3': case L'4':
281 case L'5': case L'6': case L'7': case L'8': case L'9':
282 case L':': case L';': case L'<': case L'=': case L'>':
283 case L'?':
284 case L'A': case L'B': case L'C': case L'D': case L'E':
285 case L'F': case L'G': case L'H': case L'I': case L'J':
286 case L'K': case L'L': case L'M': case L'N': case L'O':
287 case L'P': case L'Q': case L'R': case L'S': case L'T':
288 case L'U': case L'V': case L'W': case L'X': case L'Y':
289 case L'Z':
290 case L'[': case L'\\': case L']': case L'^': case L'_':
291 case L'a': case L'b': case L'c': case L'd': case L'e':
292 case L'f': case L'g': case L'h': case L'i': case L'j':
293 case L'k': case L'l': case L'm': case L'n': case L'o':
294 case L'p': case L'q': case L'r': case L's': case L't':
295 case L'u': case L'v': case L'w': case L'x': case L'y':
296 case L'z': case L'{': case L'|': case L'}': case L'~':
297 break;
298 default:
299 return (wctype_t) 0;
300 }
301# endif
302
303 /* Avoid overrunning the buffer. */
304 if (cp == s + CHAR_CLASS_MAX_LENGTH)
305 return (wctype_t) 0;
306
307 *cp++ = (char) *wcs++;
308 }
309 while (*wcs != L'\0');
310
311 *cp = '\0';
312
313# ifdef _LIBC
314 return __wctype (s);
315# else
316 return wctype (s);
317# endif
318}
319# define IS_CHAR_CLASS(string) is_char_class (string)
320
321# include "fnmatch_loop.c"
322# endif
323
324
325int
326fnmatch (const char *pattern, const char *string, int flags)
327{
328# if HANDLE_MULTIBYTE
329 if (__builtin_expect (MB_CUR_MAX, 1) != 1)
330 {
331 mbstate_t ps;
332 size_t n;
333 const char *p;
334 wchar_t *wpattern_malloc = NULL;
335 wchar_t *wpattern;
336 wchar_t *wstring_malloc = NULL;
337 wchar_t *wstring;
338 size_t alloca_used = 0;
339
340 /* Convert the strings into wide characters. */
341 memset (&ps, '\0', sizeof (ps));
342 p = pattern;
343#ifdef _LIBC
344 n = __strnlen (pattern, 1024);
345#else
346 n = strlen (pattern);
347#endif
348 if (__glibc_likely (n < 1024))
349 {
350 wpattern = (wchar_t *) alloca_account ((n + 1) * sizeof (wchar_t),
351 alloca_used);
352 n = mbsrtowcs (wpattern, &p, n + 1, &ps);
353 if (__glibc_unlikely (n == (size_t) -1))
354 /* Something wrong.
355 XXX Do we have to set `errno' to something which mbsrtows hasn't
356 already done? */
357 return -1;
358 if (p)
359 {
360 memset (&ps, '\0', sizeof (ps));
361 goto prepare_wpattern;
362 }
363 }
364 else
365 {
366 prepare_wpattern:
367 n = mbsrtowcs (NULL, &pattern, 0, &ps);
368 if (__glibc_unlikely (n == (size_t) -1))
369 /* Something wrong.
370 XXX Do we have to set `errno' to something which mbsrtows hasn't
371 already done? */
372 return -1;
373 if (__glibc_unlikely (n >= (size_t) -1 / sizeof (wchar_t)))
374 {
375 __set_errno (ENOMEM);
376 return -2;
377 }
378 wpattern_malloc = wpattern
379 = (wchar_t *) malloc ((n + 1) * sizeof (wchar_t));
380 assert (mbsinit (&ps));
381 if (wpattern == NULL)
382 return -2;
383 (void) mbsrtowcs (wpattern, &pattern, n + 1, &ps);
384 }
385
386 assert (mbsinit (&ps));
387#ifdef _LIBC
388 n = __strnlen (string, 1024);
389#else
390 n = strlen (string);
391#endif
392 p = string;
393 if (__glibc_likely (n < 1024))
394 {
395 wstring = (wchar_t *) alloca_account ((n + 1) * sizeof (wchar_t),
396 alloca_used);
397 n = mbsrtowcs (wstring, &p, n + 1, &ps);
398 if (__glibc_unlikely (n == (size_t) -1))
399 {
400 /* Something wrong.
401 XXX Do we have to set `errno' to something which
402 mbsrtows hasn't already done? */
403 free_return:
404 free (wpattern_malloc);
405 return -1;
406 }
407 if (p)
408 {
409 memset (&ps, '\0', sizeof (ps));
410 goto prepare_wstring;
411 }
412 }
413 else
414 {
415 prepare_wstring:
416 n = mbsrtowcs (NULL, &string, 0, &ps);
417 if (__glibc_unlikely (n == (size_t) -1))
418 /* Something wrong.
419 XXX Do we have to set `errno' to something which mbsrtows hasn't
420 already done? */
421 goto free_return;
422 if (__glibc_unlikely (n >= (size_t) -1 / sizeof (wchar_t)))
423 {
424 free (wpattern_malloc);
425 __set_errno (ENOMEM);
426 return -2;
427 }
428
429 wstring_malloc = wstring
430 = (wchar_t *) malloc ((n + 1) * sizeof (wchar_t));
431 if (wstring == NULL)
432 {
433 free (wpattern_malloc);
434 return -2;
435 }
436 assert (mbsinit (&ps));
437 (void) mbsrtowcs (wstring, &string, n + 1, &ps);
438 }
439
440 int res = internal_fnwmatch (wpattern, wstring, wstring + n,
441 flags & FNM_PERIOD, flags, NULL,
442 alloca_used);
443
444 free (wstring_malloc);
445 free (wpattern_malloc);
446
447 return res;
448 }
449# endif /* mbstate_t and mbsrtowcs or _LIBC. */
450
451 return internal_fnmatch (pattern, string, string + strlen (string),
452 flags & FNM_PERIOD, flags, NULL, 0);
453}
454
455# ifdef _LIBC
456# undef fnmatch
457versioned_symbol (libc, __fnmatch, fnmatch, GLIBC_2_2_3);
458# if SHLIB_COMPAT(libc, GLIBC_2_0, GLIBC_2_2_3)
459strong_alias (__fnmatch, __fnmatch_old)
460compat_symbol (libc, __fnmatch_old, fnmatch, GLIBC_2_0);
461# endif
462libc_hidden_ver (__fnmatch, fnmatch)
463# endif
464
465#endif /* _LIBC or not __GNU_LIBRARY__. */
466