1 | /* Copyright (C) 1991-2021 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 | #ifndef _LIBC |
19 | # include <libc-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 <fnmatch.h> |
28 | |
29 | #include <assert.h> |
30 | #include <errno.h> |
31 | #include <ctype.h> |
32 | #include <string.h> |
33 | #include <stdlib.h> |
34 | #if defined _LIBC || HAVE_ALLOCA |
35 | # include <alloca.h> |
36 | #endif |
37 | #include <wchar.h> |
38 | #include <wctype.h> |
39 | #include <stddef.h> |
40 | #include <stdbool.h> |
41 | |
42 | /* We need some of the locale data (the collation sequence information) |
43 | but there is no interface to get this information in general. Therefore |
44 | we support a correct implementation only in glibc. */ |
45 | #ifdef _LIBC |
46 | # include "../locale/localeinfo.h" |
47 | # include "../locale/coll-lookup.h" |
48 | # include <shlib-compat.h> |
49 | |
50 | # define CONCAT(a,b) __CONCAT(a,b) |
51 | # define btowc __btowc |
52 | # define iswctype __iswctype |
53 | # define mbsrtowcs __mbsrtowcs |
54 | # define mempcpy __mempcpy |
55 | # define strnlen __strnlen |
56 | # define towlower __towlower |
57 | # define wcscat __wcscat |
58 | # define wcslen __wcslen |
59 | # define wctype __wctype |
60 | # define wmemchr __wmemchr |
61 | # define wmempcpy __wmempcpy |
62 | # define fnmatch __fnmatch |
63 | extern int fnmatch (const char *pattern, const char *string, int flags); |
64 | #endif |
65 | |
66 | #ifdef _LIBC |
67 | # if __GNUC__ >= 7 |
68 | # define FALLTHROUGH __attribute__ ((__fallthrough__)) |
69 | # else |
70 | # define FALLTHROUGH ((void) 0) |
71 | # endif |
72 | #else |
73 | # include "attribute.h" |
74 | #endif |
75 | |
76 | #include <intprops.h> |
77 | #include <flexmember.h> |
78 | #include <scratch_buffer.h> |
79 | |
80 | #ifdef _LIBC |
81 | typedef ptrdiff_t idx_t; |
82 | #else |
83 | # include "idx.h" |
84 | #endif |
85 | |
86 | /* We often have to test for FNM_FILE_NAME and FNM_PERIOD being both set. */ |
87 | #define NO_LEADING_PERIOD(flags) \ |
88 | ((flags & (FNM_FILE_NAME | FNM_PERIOD)) == (FNM_FILE_NAME | FNM_PERIOD)) |
89 | |
90 | #ifndef _LIBC |
91 | # if HAVE_ALLOCA |
92 | /* The OS usually guarantees only one guard page at the bottom of the stack, |
93 | and a page size can be as small as 4096 bytes. So we cannot safely |
94 | allocate anything larger than 4096 bytes. Also care for the possibility |
95 | of a few compiler-allocated temporary stack slots. */ |
96 | # define __libc_use_alloca(n) ((n) < 4032) |
97 | # else |
98 | /* Just use malloc. */ |
99 | # define __libc_use_alloca(n) false |
100 | # undef alloca |
101 | # define alloca(n) malloc (n) |
102 | # endif |
103 | # define alloca_account(size, avar) ((avar) += (size), alloca (size)) |
104 | #endif |
105 | |
106 | /* Provide support for user-defined character classes, based on the functions |
107 | from ISO C 90 amendment 1. */ |
108 | #ifdef CHARCLASS_NAME_MAX |
109 | # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX |
110 | #else |
111 | /* This shouldn't happen but some implementation might still have this |
112 | problem. Use a reasonable default value. */ |
113 | # define CHAR_CLASS_MAX_LENGTH 256 |
114 | #endif |
115 | |
116 | #define IS_CHAR_CLASS(string) wctype (string) |
117 | |
118 | /* Avoid depending on library functions or files |
119 | whose names are inconsistent. */ |
120 | |
121 | /* Global variable. */ |
122 | static int posixly_correct; |
123 | |
124 | /* Note that this evaluates C many times. */ |
125 | #define FOLD(c) ((flags & FNM_CASEFOLD) ? tolower (c) : (c)) |
126 | #define CHAR char |
127 | #define UCHAR unsigned char |
128 | #define INT int |
129 | #define FCT internal_fnmatch |
130 | #define EXT ext_match |
131 | #define END end_pattern |
132 | #define STRUCT fnmatch_struct |
133 | #define L_(CS) CS |
134 | #define BTOWC(C) btowc (C) |
135 | #define STRLEN(S) strlen (S) |
136 | #define STRCAT(D, S) strcat (D, S) |
137 | #define MEMPCPY(D, S, N) mempcpy (D, S, N) |
138 | #define MEMCHR(S, C, N) memchr (S, C, N) |
139 | #define WIDE_CHAR_VERSION 0 |
140 | #ifdef _LIBC |
141 | # include <locale/weight.h> |
142 | # define FINDIDX findidx |
143 | #endif |
144 | #include "fnmatch_loop.c" |
145 | |
146 | |
147 | #define FOLD(c) ((flags & FNM_CASEFOLD) ? towlower (c) : (c)) |
148 | #define CHAR wchar_t |
149 | #define UCHAR wint_t |
150 | #define INT wint_t |
151 | #define FCT internal_fnwmatch |
152 | #define EXT ext_wmatch |
153 | #define END end_wpattern |
154 | #define L_(CS) L##CS |
155 | #define BTOWC(C) (C) |
156 | #define STRLEN(S) wcslen (S) |
157 | #define STRCAT(D, S) wcscat (D, S) |
158 | #define MEMPCPY(D, S, N) wmempcpy (D, S, N) |
159 | #define MEMCHR(S, C, N) wmemchr (S, C, N) |
160 | #define WIDE_CHAR_VERSION 1 |
161 | #ifdef _LIBC |
162 | /* Change the name the header defines so it doesn't conflict with |
163 | the <locale/weight.h> version included above. */ |
164 | # define findidx findidxwc |
165 | # include <locale/weightwc.h> |
166 | # undef findidx |
167 | # define FINDIDX findidxwc |
168 | #endif |
169 | |
170 | #undef IS_CHAR_CLASS |
171 | /* We have to convert the wide character string in a multibyte string. But |
172 | we know that the character class names consist of alphanumeric characters |
173 | from the portable character set, and since the wide character encoding |
174 | for a member of the portable character set is the same code point as |
175 | its single-byte encoding, we can use a simplified method to convert the |
176 | string to a multibyte character string. */ |
177 | static wctype_t |
178 | is_char_class (const wchar_t *wcs) |
179 | { |
180 | char s[CHAR_CLASS_MAX_LENGTH + 1]; |
181 | char *cp = s; |
182 | |
183 | do |
184 | { |
185 | /* Test for a printable character from the portable character set. */ |
186 | #ifdef _LIBC |
187 | if (*wcs < 0x20 || *wcs > 0x7e |
188 | || *wcs == 0x24 || *wcs == 0x40 || *wcs == 0x60) |
189 | return (wctype_t) 0; |
190 | #else |
191 | switch (*wcs) |
192 | { |
193 | case L' ': case L'!': case L'"': case L'#': case L'%': |
194 | case L'&': case L'\'': case L'(': case L')': case L'*': |
195 | case L'+': case L',': case L'-': case L'.': case L'/': |
196 | case L'0': case L'1': case L'2': case L'3': case L'4': |
197 | case L'5': case L'6': case L'7': case L'8': case L'9': |
198 | case L':': case L';': case L'<': case L'=': case L'>': |
199 | case L'?': |
200 | case L'A': case L'B': case L'C': case L'D': case L'E': |
201 | case L'F': case L'G': case L'H': case L'I': case L'J': |
202 | case L'K': case L'L': case L'M': case L'N': case L'O': |
203 | case L'P': case L'Q': case L'R': case L'S': case L'T': |
204 | case L'U': case L'V': case L'W': case L'X': case L'Y': |
205 | case L'Z': |
206 | case L'[': case L'\\': case L']': case L'^': case L'_': |
207 | case L'a': case L'b': case L'c': case L'd': case L'e': |
208 | case L'f': case L'g': case L'h': case L'i': case L'j': |
209 | case L'k': case L'l': case L'm': case L'n': case L'o': |
210 | case L'p': case L'q': case L'r': case L's': case L't': |
211 | case L'u': case L'v': case L'w': case L'x': case L'y': |
212 | case L'z': case L'{': case L'|': case L'}': case L'~': |
213 | break; |
214 | default: |
215 | return (wctype_t) 0; |
216 | } |
217 | #endif |
218 | |
219 | /* Avoid overrunning the buffer. */ |
220 | if (cp == s + CHAR_CLASS_MAX_LENGTH) |
221 | return (wctype_t) 0; |
222 | |
223 | *cp++ = (char) *wcs++; |
224 | } |
225 | while (*wcs != L'\0'); |
226 | |
227 | *cp = '\0'; |
228 | |
229 | return wctype (s); |
230 | } |
231 | #define IS_CHAR_CLASS(string) is_char_class (string) |
232 | |
233 | #include "fnmatch_loop.c" |
234 | |
235 | static int |
236 | fnmatch_convert_to_wide (const char *str, struct scratch_buffer *buf, |
237 | size_t *n) |
238 | { |
239 | mbstate_t ps; |
240 | memset (&ps, '\0', sizeof (ps)); |
241 | |
242 | size_t nw = buf->length / sizeof (wchar_t); |
243 | *n = strnlen (str, nw - 1); |
244 | if (__glibc_likely (*n < nw)) |
245 | { |
246 | const char *p = str; |
247 | *n = mbsrtowcs (buf->data, &p, *n + 1, &ps); |
248 | if (__glibc_unlikely (*n == (size_t) -1)) |
249 | /* Something wrong. |
250 | XXX Do we have to set 'errno' to something which mbsrtows hasn't |
251 | already done? */ |
252 | return -1; |
253 | if (p == NULL) |
254 | return 0; |
255 | memset (&ps, '\0', sizeof (ps)); |
256 | } |
257 | |
258 | *n = mbsrtowcs (NULL, &str, 0, &ps); |
259 | if (__glibc_unlikely (*n == (size_t) -1)) |
260 | return -1; |
261 | if (!scratch_buffer_set_array_size (buf, *n + 1, sizeof (wchar_t))) |
262 | { |
263 | __set_errno (ENOMEM); |
264 | return -2; |
265 | } |
266 | assert (mbsinit (&ps)); |
267 | mbsrtowcs (buf->data, &str, *n + 1, &ps); |
268 | return 0; |
269 | } |
270 | |
271 | int |
272 | fnmatch (const char *pattern, const char *string, int flags) |
273 | { |
274 | if (__glibc_unlikely (MB_CUR_MAX != 1)) |
275 | { |
276 | size_t n; |
277 | struct scratch_buffer wpattern; |
278 | scratch_buffer_init (&wpattern); |
279 | struct scratch_buffer wstring; |
280 | scratch_buffer_init (&wstring); |
281 | int r; |
282 | |
283 | /* Convert the strings into wide characters. Any conversion issue |
284 | fallback to the ascii version. */ |
285 | r = fnmatch_convert_to_wide (pattern, &wpattern, &n); |
286 | if (r == 0) |
287 | { |
288 | r = fnmatch_convert_to_wide (string, &wstring, &n); |
289 | if (r == 0) |
290 | r = internal_fnwmatch (wpattern.data, wstring.data, |
291 | (wchar_t *) wstring.data + n, |
292 | flags & FNM_PERIOD, flags, NULL, |
293 | false); |
294 | } |
295 | |
296 | scratch_buffer_free (&wstring); |
297 | scratch_buffer_free (&wpattern); |
298 | |
299 | if (r == -2 || r == 0) |
300 | return r; |
301 | } |
302 | |
303 | return internal_fnmatch (pattern, string, string + strlen (string), |
304 | flags & FNM_PERIOD, flags, NULL, 0); |
305 | } |
306 | |
307 | #undef fnmatch |
308 | versioned_symbol (libc, __fnmatch, fnmatch, GLIBC_2_2_3); |
309 | #if SHLIB_COMPAT(libc, GLIBC_2_0, GLIBC_2_2_3) |
310 | strong_alias (__fnmatch, __fnmatch_old) |
311 | compat_symbol (libc, __fnmatch_old, fnmatch, GLIBC_2_0); |
312 | #endif |
313 | libc_hidden_ver (__fnmatch, fnmatch) |
314 | |