1/* Copyright (C) 1991-2020 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 <https://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/coll-lookup.h"
57# include <shlib-compat.h>
58
59# define CONCAT(a,b) __CONCAT(a,b)
60# define mbsrtowcs __mbsrtowcs
61# define fnmatch __fnmatch
62extern int fnmatch (const char *pattern, const char *string, int flags);
63#endif
64
65/* We often have to test for FNM_FILE_NAME and FNM_PERIOD being both set. */
66#define NO_LEADING_PERIOD(flags) \
67 ((flags & (FNM_FILE_NAME | FNM_PERIOD)) == (FNM_FILE_NAME | FNM_PERIOD))
68
69/* Comment out all this code if we are using the GNU C Library, and are not
70 actually compiling the library itself. This code is part of the GNU C
71 Library, but also included in many other GNU distributions. Compiling
72 and linking in this code is a waste when using the GNU C library
73 (especially if it is a shared library). Rather than having every GNU
74 program understand `configure --with-gnu-libc' and omit the object files,
75 it is simpler to just do this in the source for each such file. */
76
77#if defined _LIBC || !defined __GNU_LIBRARY__
78
79
80# if defined STDC_HEADERS || !defined isascii
81# define ISASCII(c) 1
82# else
83# define ISASCII(c) isascii(c)
84# endif
85
86# ifdef isblank
87# define ISBLANK(c) (ISASCII (c) && isblank (c))
88# else
89# define ISBLANK(c) ((c) == ' ' || (c) == '\t')
90# endif
91# ifdef isgraph
92# define ISGRAPH(c) (ISASCII (c) && isgraph (c))
93# else
94# define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
95# endif
96
97# define ISPRINT(c) (ISASCII (c) && isprint (c))
98# define ISDIGIT(c) (ISASCII (c) && isdigit (c))
99# define ISALNUM(c) (ISASCII (c) && isalnum (c))
100# define ISALPHA(c) (ISASCII (c) && isalpha (c))
101# define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
102# define ISLOWER(c) (ISASCII (c) && islower (c))
103# define ISPUNCT(c) (ISASCII (c) && ispunct (c))
104# define ISSPACE(c) (ISASCII (c) && isspace (c))
105# define ISUPPER(c) (ISASCII (c) && isupper (c))
106# define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
107
108# define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
109
110# if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
111/* The GNU C library provides support for user-defined character classes
112 and the functions from ISO C amendement 1. */
113# ifdef CHARCLASS_NAME_MAX
114# define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
115# else
116/* This shouldn't happen but some implementation might still have this
117 problem. Use a reasonable default value. */
118# define CHAR_CLASS_MAX_LENGTH 256
119# endif
120
121# ifdef _LIBC
122# define IS_CHAR_CLASS(string) __wctype (string)
123# else
124# define IS_CHAR_CLASS(string) wctype (string)
125# endif
126
127# ifdef _LIBC
128# define ISWCTYPE(WC, WT) __iswctype (WC, WT)
129# else
130# define ISWCTYPE(WC, WT) iswctype (WC, WT)
131# endif
132
133# if (HAVE_MBSTATE_T && HAVE_MBSRTOWCS) || _LIBC
134/* In this case we are implementing the multibyte character handling. */
135# define HANDLE_MULTIBYTE 1
136# endif
137
138# else
139# define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
140
141# define IS_CHAR_CLASS(string) \
142 (STREQ (string, "alpha") || STREQ (string, "upper") \
143 || STREQ (string, "lower") || STREQ (string, "digit") \
144 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
145 || STREQ (string, "space") || STREQ (string, "print") \
146 || STREQ (string, "punct") || STREQ (string, "graph") \
147 || STREQ (string, "cntrl") || STREQ (string, "blank"))
148# endif
149
150/* Avoid depending on library functions or files
151 whose names are inconsistent. */
152
153# if !defined _LIBC && !defined getenv
154extern char *getenv ();
155# endif
156
157# ifndef errno
158extern int errno;
159# endif
160
161/* Global variable. */
162static int posixly_correct;
163
164/* This function doesn't exist on most systems. */
165
166# if !defined HAVE___STRCHRNUL && !defined _LIBC
167static char *
168__strchrnul (const char *s, int c)
169{
170 char *result = strchr (s, c);
171 if (result == NULL)
172 result = strchr (s, '\0');
173 return result;
174}
175# endif
176
177# if HANDLE_MULTIBYTE && !defined HAVE___STRCHRNUL && !defined _LIBC
178static wchar_t *
179__wcschrnul (const wchar_t *s, wint_t c)
180{
181 wchar_t *result = wcschr (s, c);
182 if (result == NULL)
183 result = wcschr (s, '\0');
184 return result;
185}
186# endif
187
188/* Note that this evaluates C many times. */
189# ifdef _LIBC
190# define FOLD(c) ((flags & FNM_CASEFOLD) ? tolower (c) : (c))
191# else
192# define FOLD(c) ((flags & FNM_CASEFOLD) && ISUPPER (c) ? tolower (c) : (c))
193# endif
194# define CHAR char
195# define UCHAR unsigned char
196# define INT int
197# define FCT internal_fnmatch
198# define EXT ext_match
199# define END end_pattern
200# define STRUCT fnmatch_struct
201# define L(CS) CS
202# ifdef _LIBC
203# define BTOWC(C) __btowc (C)
204# else
205# define BTOWC(C) btowc (C)
206# endif
207# define STRLEN(S) strlen (S)
208# define STRCAT(D, S) strcat (D, S)
209# define MEMPCPY(D, S, N) __mempcpy (D, S, N)
210# define MEMCHR(S, C, N) memchr (S, C, N)
211# define STRCOLL(S1, S2) strcoll (S1, S2)
212# define WIDE_CHAR_VERSION 0
213# include <locale/weight.h>
214# define FINDIDX findidx
215# include "fnmatch_loop.c"
216
217
218# if HANDLE_MULTIBYTE
219/* Note that this evaluates C many times. */
220# ifdef _LIBC
221# define FOLD(c) ((flags & FNM_CASEFOLD) ? __towlower (c) : (c))
222# else
223# define FOLD(c) ((flags & FNM_CASEFOLD) && ISUPPER (c) ? towlower (c) : (c))
224# endif
225# define CHAR wchar_t
226# define UCHAR wint_t
227# define INT wint_t
228# define FCT internal_fnwmatch
229# define EXT ext_wmatch
230# define END end_wpattern
231# define STRUCT fnwmatch_struct
232# define L(CS) L##CS
233# define BTOWC(C) (C)
234# define STRLEN(S) __wcslen (S)
235# define STRCAT(D, S) __wcscat (D, S)
236# define MEMPCPY(D, S, N) __wmempcpy (D, S, N)
237# define MEMCHR(S, C, N) __wmemchr (S, C, N)
238# define STRCOLL(S1, S2) wcscoll (S1, S2)
239# ifdef _LIBC
240# define WMEMCMP(S1, S2, N) __wmemcmp (S1, S2, N)
241# else
242# define WMEMCMP(S1, S2, N) wmemcmp (S1, S2, N)
243# endif
244# define WIDE_CHAR_VERSION 1
245/* Change the name the header defines so it doesn't conflict with
246 the <locale/weight.h> version included above. */
247# define findidx findidxwc
248# include <locale/weightwc.h>
249# undef findidx
250# define FINDIDX findidxwc
251
252# undef IS_CHAR_CLASS
253/* We have to convert the wide character string in a multibyte string. But
254 we know that the character class names consist of alphanumeric characters
255 from the portable character set, and since the wide character encoding
256 for a member of the portable character set is the same code point as
257 its single-byte encoding, we can use a simplified method to convert the
258 string to a multibyte character string. */
259static wctype_t
260is_char_class (const wchar_t *wcs)
261{
262 char s[CHAR_CLASS_MAX_LENGTH + 1];
263 char *cp = s;
264
265 do
266 {
267 /* Test for a printable character from the portable character set. */
268# ifdef _LIBC
269 if (*wcs < 0x20 || *wcs > 0x7e
270 || *wcs == 0x24 || *wcs == 0x40 || *wcs == 0x60)
271 return (wctype_t) 0;
272# else
273 switch (*wcs)
274 {
275 case L' ': case L'!': case L'"': case L'#': case L'%':
276 case L'&': case L'\'': case L'(': case L')': case L'*':
277 case L'+': case L',': case L'-': case L'.': case L'/':
278 case L'0': case L'1': case L'2': case L'3': case L'4':
279 case L'5': case L'6': case L'7': case L'8': case L'9':
280 case L':': case L';': case L'<': case L'=': case L'>':
281 case L'?':
282 case L'A': case L'B': case L'C': case L'D': case L'E':
283 case L'F': case L'G': case L'H': case L'I': case L'J':
284 case L'K': case L'L': case L'M': case L'N': case L'O':
285 case L'P': case L'Q': case L'R': case L'S': case L'T':
286 case L'U': case L'V': case L'W': case L'X': case L'Y':
287 case L'Z':
288 case L'[': case L'\\': case L']': case L'^': case L'_':
289 case L'a': case L'b': case L'c': case L'd': case L'e':
290 case L'f': case L'g': case L'h': case L'i': case L'j':
291 case L'k': case L'l': case L'm': case L'n': case L'o':
292 case L'p': case L'q': case L'r': case L's': case L't':
293 case L'u': case L'v': case L'w': case L'x': case L'y':
294 case L'z': case L'{': case L'|': case L'}': case L'~':
295 break;
296 default:
297 return (wctype_t) 0;
298 }
299# endif
300
301 /* Avoid overrunning the buffer. */
302 if (cp == s + CHAR_CLASS_MAX_LENGTH)
303 return (wctype_t) 0;
304
305 *cp++ = (char) *wcs++;
306 }
307 while (*wcs != L'\0');
308
309 *cp = '\0';
310
311# ifdef _LIBC
312 return __wctype (s);
313# else
314 return wctype (s);
315# endif
316}
317# define IS_CHAR_CLASS(string) is_char_class (string)
318
319# include "fnmatch_loop.c"
320# endif
321
322
323int
324fnmatch (const char *pattern, const char *string, int flags)
325{
326# if HANDLE_MULTIBYTE
327 if (__builtin_expect (MB_CUR_MAX, 1) != 1)
328 {
329 mbstate_t ps;
330 size_t n;
331 const char *p;
332 wchar_t *wpattern_malloc = NULL;
333 wchar_t *wpattern;
334 wchar_t *wstring_malloc = NULL;
335 wchar_t *wstring;
336 size_t alloca_used = 0;
337
338 /* Convert the strings into wide characters. */
339 memset (&ps, '\0', sizeof (ps));
340 p = pattern;
341#ifdef _LIBC
342 n = __strnlen (pattern, 1024);
343#else
344 n = strlen (pattern);
345#endif
346 if (__glibc_likely (n < 1024))
347 {
348 wpattern = (wchar_t *) alloca_account ((n + 1) * sizeof (wchar_t),
349 alloca_used);
350 n = mbsrtowcs (wpattern, &p, n + 1, &ps);
351 if (__glibc_unlikely (n == (size_t) -1))
352 /* Something wrong.
353 XXX Do we have to set `errno' to something which mbsrtows hasn't
354 already done? */
355 return -1;
356 if (p)
357 {
358 memset (&ps, '\0', sizeof (ps));
359 goto prepare_wpattern;
360 }
361 }
362 else
363 {
364 prepare_wpattern:
365 n = mbsrtowcs (NULL, &pattern, 0, &ps);
366 if (__glibc_unlikely (n == (size_t) -1))
367 /* Something wrong.
368 XXX Do we have to set `errno' to something which mbsrtows hasn't
369 already done? */
370 return -1;
371 if (__glibc_unlikely (n >= (size_t) -1 / sizeof (wchar_t)))
372 {
373 __set_errno (ENOMEM);
374 return -2;
375 }
376 wpattern_malloc = wpattern
377 = (wchar_t *) malloc ((n + 1) * sizeof (wchar_t));
378 assert (mbsinit (&ps));
379 if (wpattern == NULL)
380 return -2;
381 (void) mbsrtowcs (wpattern, &pattern, n + 1, &ps);
382 }
383
384 assert (mbsinit (&ps));
385#ifdef _LIBC
386 n = __strnlen (string, 1024);
387#else
388 n = strlen (string);
389#endif
390 p = string;
391 if (__glibc_likely (n < 1024))
392 {
393 wstring = (wchar_t *) alloca_account ((n + 1) * sizeof (wchar_t),
394 alloca_used);
395 n = mbsrtowcs (wstring, &p, n + 1, &ps);
396 if (__glibc_unlikely (n == (size_t) -1))
397 {
398 /* Something wrong.
399 XXX Do we have to set `errno' to something which
400 mbsrtows hasn't already done? */
401 free_return:
402 free (wpattern_malloc);
403 return -1;
404 }
405 if (p)
406 {
407 memset (&ps, '\0', sizeof (ps));
408 goto prepare_wstring;
409 }
410 }
411 else
412 {
413 prepare_wstring:
414 n = mbsrtowcs (NULL, &string, 0, &ps);
415 if (__glibc_unlikely (n == (size_t) -1))
416 /* Something wrong.
417 XXX Do we have to set `errno' to something which mbsrtows hasn't
418 already done? */
419 goto free_return;
420 if (__glibc_unlikely (n >= (size_t) -1 / sizeof (wchar_t)))
421 {
422 free (wpattern_malloc);
423 __set_errno (ENOMEM);
424 return -2;
425 }
426
427 wstring_malloc = wstring
428 = (wchar_t *) malloc ((n + 1) * sizeof (wchar_t));
429 if (wstring == NULL)
430 {
431 free (wpattern_malloc);
432 return -2;
433 }
434 assert (mbsinit (&ps));
435 (void) mbsrtowcs (wstring, &string, n + 1, &ps);
436 }
437
438 int res = internal_fnwmatch (wpattern, wstring, wstring + n,
439 flags & FNM_PERIOD, flags, NULL,
440 alloca_used);
441
442 free (wstring_malloc);
443 free (wpattern_malloc);
444
445 return res;
446 }
447# endif /* mbstate_t and mbsrtowcs or _LIBC. */
448
449 return internal_fnmatch (pattern, string, string + strlen (string),
450 flags & FNM_PERIOD, flags, NULL, 0);
451}
452
453# ifdef _LIBC
454# undef fnmatch
455versioned_symbol (libc, __fnmatch, fnmatch, GLIBC_2_2_3);
456# if SHLIB_COMPAT(libc, GLIBC_2_0, GLIBC_2_2_3)
457strong_alias (__fnmatch, __fnmatch_old)
458compat_symbol (libc, __fnmatch_old, fnmatch, GLIBC_2_0);
459# endif
460libc_hidden_ver (__fnmatch, fnmatch)
461# endif
462
463#endif /* _LIBC or not __GNU_LIBRARY__. */
464