1 /* Test for UTF-8 regular expression optimizations.
2 Copyright (C) 2003-2023 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4
5 The GNU C Library is free software; you can redistribute it and/or
6 modify it under the terms of the GNU Lesser General Public
7 License as published by the Free Software Foundation; either
8 version 2.1 of the License, or (at your option) any later version.
9
10 The GNU C Library is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 Lesser General Public License for more details.
14
15 You should have received a copy of the GNU Lesser General Public
16 License along with the GNU C Library; if not, see
17 <https://www.gnu.org/licenses/>. */
18
19 #include <sys/types.h>
20 #include <mcheck.h>
21 #include <regex.h>
22 #include <stdio.h>
23 #include <stdlib.h>
24 #include <string.h>
25 #include <locale.h>
26
27 #define RE_NO_INTERNAL_PROTOTYPES 1
28 #include "regex_internal.h"
29
30 #define BRE RE_SYNTAX_POSIX_BASIC
31 #define ERE RE_SYNTAX_POSIX_EXTENDED
32
33 static struct
34 {
35 int syntax;
36 const char *pattern;
37 const char *string;
38 int res, optimize;
39 } tests[] = {
40 /* \xc3\x84 LATIN CAPITAL LETTER A WITH DIAERESIS
41 \xc3\x96 LATIN CAPITAL LETTER O WITH DIAERESIS
42 \xc3\xa4 LATIN SMALL LETTER A WITH DIAERESIS
43 \xc3\xb6 LATIN SMALL LETTER O WITH DIAERESIS
44 \xe2\x80\x94 EM DASH */
45 /* Should be optimized. */
46 {BRE, "foo", "b\xc3\xa4rfoob\xc3\xa4z", 4, 1},
47 {BRE, "b\xc3\xa4z", "b\xc3\xa4rfoob\xc3\xa4z", 7, 1},
48 {BRE, "b\xc3\xa4*z", "b\xc3\xa4rfoob\xc3\xa4z", 7, 1},
49 {BRE, "b\xc3\xa4*z", "b\xc3\xa4rfoobz", 7, 1},
50 {BRE, "b\xc3\xa4\\+z", "b\xc3\xa4rfoob\xc3\xa4\xc3\xa4z", 7, 1},
51 {BRE, "b\xc3\xa4\\?z", "b\xc3\xa4rfoob\xc3\xa4z", 7, 1},
52 {BRE, "b\xc3\xa4\\{1,2\\}z", "b\xc3\xa4rfoob\xc3\xa4z", 7, 1},
53 {BRE, "^x\\|xy*z$", "\xc3\xb6xyyz", 2, 1},
54 {BRE, "^x\\\\y\\{6\\}z\\+", "x\\yyyyyyzz\xc3\xb6", 0, 1},
55 {BRE, "^x\\\\y\\{2,36\\}z\\+", "x\\yzz\xc3\xb6", -1, 1},
56 {BRE, "^x\\\\y\\{,3\\}z\\+", "x\\yyyzz\xc3\xb6", 0, 1},
57 {BRE, "^x\\|x\xc3\xa4*z$", "\xc3\xb6x\xc3\xa4\xc3\xa4z", 2, 1},
58 {BRE, "^x\\\\\xc3\x84\\{6\\}z\\+",
59 "x\\\xc3\x84\xc3\x84\xc3\x84\xc3\x84\xc3\x84\xc3\x84zz\xc3\xb6", 0, 1},
60 {BRE, "^x\\\\\xc3\x84\\{2,36\\}z\\+", "x\\\xc3\x84zz\xc3\xb6", -1, 1},
61 {BRE, "^x\\\\\xc3\x84\\{,3\\}z\\+",
62 "x\\\xc3\x84\xc3\x84\xc3\x84zz\xc3\xb6", 0, 1},
63 {BRE, "x[C]y", "axCy", 1, 1},
64 {BRE, "x[ABC]y", "axCy", 1, 1},
65 {BRE, "\\`x\\|z\\'", "x\xe2\x80\x94", 0, 1},
66 {BRE, "\\(xy\\)z\\1a\\1", "\xe2\x80\x94xyzxyaxy\xc3\x84", 3, 1},
67 {BRE, "xy\\?z", "\xc3\x84xz\xc3\xb6", 2, 1},
68 {BRE, "\\`\xc3\x84\\|z\\'", "\xc3\x84\xe2\x80\x94", 0, 1},
69 {BRE, "\\(x\xc3\x84\\)z\\1\x61\\1",
70 "\xe2\x80\x94x\xc3\x84zx\xc3\x84\x61x\xc3\x84\xc3\x96", 3, 1},
71 {BRE, "x\xc3\x96\\?z", "\xc3\x84xz\xc3\xb6", 2, 1},
72 {BRE, "x.y", "ax\xe2\x80\x94yz", 1, 1},
73 {BRE, "x.*z", "\xc3\x84xz", 2, 1},
74 {BRE, "x.*z", "\xc3\x84x\xe2\x80\x94z", 2, 1},
75 {BRE, "x.*z", "\xc3\x84x\xe2\x80\x94y\xf1\x90\x80\x90z", 2, 1},
76 {BRE, "x.*z", "\xc3\x84x\xe2\x80\x94\xc3\x94\xf1\x90\x80\x90z", 2, 1},
77 {BRE, "x.\\?z", "axz", 1, 1},
78 {BRE, "x.\\?z", "axyz", 1, 1},
79 {BRE, "x.\\?z", "ax\xc3\x84z", 1, 1},
80 {BRE, "x.\\?z", "ax\xe2\x80\x94z", 1, 1},
81 {BRE, "x.\\?z", "ax\xf0\x9d\x80\x80z", 1, 1},
82 {BRE, "x.\\?z", "ax\xf9\x81\x82\x83\x84z", 1, 1},
83 {BRE, "x.\\?z", "ax\xfd\xbf\xbf\xbf\xbf\xbfz", 1, 1},
84 {BRE, ".", "y", 0, 1},
85 {BRE, ".", "\xc3\x84", 0, 1},
86 {BRE, ".", "\xe2\x80\x94", 0, 1},
87 {BRE, ".", "\xf0\x9d\x80\x80", 0, 1},
88 {BRE, ".", "\xf9\x81\x82\x83\x84", 0, 1},
89 {BRE, ".", "\xfd\xbf\xbf\xbf\xbf\xbf", 0, 1},
90 {BRE, "x.\\?z", "axyyz", -1, 1},
91 {BRE, "x.\\?z", "ax\xc3\x84\xc3\x96z", -1, 1},
92 {BRE, "x.\\?z", "ax\xe2\x80\x94\xc3\xa4z", -1, 1},
93 {BRE, "x.\\?z", "ax\xf0\x9d\x80\x80yz", -1, 1},
94 {BRE, "x.\\?z", "ax\xf9\x81\x82\x83\x84\xf0\x9d\x80\x81z", -1, 1},
95 {BRE, "x.\\?z", "ax\xfd\xbf\xbf\xbf\xbf\xbf\xc3\x96z", -1, 1},
96 {BRE, "x.\\+z", "\xe2\x80\x94xz", -1, 1},
97 {BRE, "x.\\+z", "\xe2\x80\x94xyz", 3, 1},
98 {BRE, "x.\\+z", "\xe2\x80\x94x\xc3\x84y\xe2\x80\x94z", 3, 1},
99 {BRE, "x.\\+z", "\xe2\x80\x94x\xe2\x80\x94z", 3, 1},
100 {BRE, "x.\\+z", "\xe2\x80\x94x\xf0\x9d\x80\x80\xc3\x84z", 3, 1},
101 {BRE, "x.\\+z", "\xe2\x80\x94x.~\xe2\x80\x94\xf9\x81\x82\x83\x84z", 3, 1},
102 {BRE, "x.\\+z", "\xe2\x80\x94x\xfd\xbf\xbf\xbf\xbf\xbfz", 3, 1},
103 {BRE, "x.\\{1,2\\}z", "\xe2\x80\x94xz", -1, 1},
104 {BRE, "x.\\{1,2\\}z", "\xe2\x80\x94x\xc3\x96y\xc3\xa4z", -1, 1},
105 {BRE, "x.\\{1,2\\}z", "\xe2\x80\x94xyz", 3, 1},
106 {BRE, "x.\\{1,2\\}z", "\xe2\x80\x94x\xc3\x84\xe2\x80\x94z", 3, 1},
107 {BRE, "x.\\{1,2\\}z", "\xe2\x80\x94x\xe2\x80\x94z", 3, 1},
108 {BRE, "x.\\{1,2\\}z", "\xe2\x80\x94x\xf0\x9d\x80\x80\xc3\x84z", 3, 1},
109 {BRE, "x.\\{1,2\\}z", "\xe2\x80\x94x~\xe2\x80\x94z", 3, 1},
110 {BRE, "x.\\{1,2\\}z", "\xe2\x80\x94x\xfd\xbf\xbf\xbf\xbf\xbfz", 3, 1},
111 {BRE, "x\\(.w\\|\xc3\x86\\)\\?z", "axz", 1, 1},
112 {BRE, "x\\(.w\\|\xc3\x86\\)\\?z", "ax\xfd\xbf\xbf\xbf\xbf\xbfwz", 1, 1},
113 {BRE, "x\\(.w\\|\xc3\x86\\)\\?z", "ax\xc3\x86z", 1, 1},
114 {BRE, "x\\(.w\\|\xc3\x86\\)\\?z", "ax\xe2\x80\x96wz", 1, 1},
115 {ERE, "foo", "b\xc3\xa4rfoob\xc3\xa4z", 4, 1},
116 {ERE, "^x|xy*z$", "\xc3\xb6xyyz", 2, 1},
117 {ERE, "^x\\\\y{6}z+", "x\\yyyyyyzz\xc3\xb6", 0, 1},
118 {ERE, "^x\\\\y{2,36}z+", "x\\yzz\xc3\xb6", -1, 1},
119 {ERE, "^x\\\\y{,3}z+", "x\\yyyzz\xc3\xb6", 0, 1},
120 {ERE, "x[C]y", "axCy", 1, 1},
121 {ERE, "x[ABC]y", "axCy", 1, 1},
122 {ERE, "\\`x|z\\'", "x\xe2\x80\x94", 0, 1},
123 {ERE, "(xy)z\\1a\\1", "\xe2\x80\x94xyzxyaxy\xc3\x84", 3, 1},
124 {ERE, "xy?z", "\xc3\x84xz\xc3\xb6", 2, 1},
125 {ERE, "x.y", "ax\xe2\x80\x94yz", 1, 1},
126 {ERE, "x.*z", "\xc3\x84xz", 2, 1},
127 {ERE, "x.*z", "\xc3\x84x\xe2\x80\x94z", 2, 1},
128 {ERE, "x.*z", "\xc3\x84x\xe2\x80\x94y\xf1\x90\x80\x90z", 2, 1},
129 {ERE, "x.*z", "\xc3\x84x\xe2\x80\x94\xc3\x94\xf1\x90\x80\x90z", 2, 1},
130 {ERE, "x.?z", "axz", 1, 1},
131 {ERE, "x.?z", "axyz", 1, 1},
132 {ERE, "x.?z", "ax\xc3\x84z", 1, 1},
133 {ERE, "x.?z", "ax\xe2\x80\x94z", 1, 1},
134 {ERE, "x.?z", "ax\xf0\x9d\x80\x80z", 1, 1},
135 {ERE, "x.?z", "ax\xf9\x81\x82\x83\x84z", 1, 1},
136 {ERE, "x.?z", "ax\xfd\xbf\xbf\xbf\xbf\xbfz", 1, 1},
137 {ERE, "x.?z", "axyyz", -1, 1},
138 {ERE, "x.?z", "ax\xc3\x84\xc3\x96z", -1, 1},
139 {ERE, "x.?z", "ax\xe2\x80\x94\xc3\xa4z", -1, 1},
140 {ERE, "x.?z", "ax\xf0\x9d\x80\x80yz", -1, 1},
141 {ERE, "x.?z", "ax\xf9\x81\x82\x83\x84\xf0\x9d\x80\x81z", -1, 1},
142 {ERE, "x.?z", "ax\xfd\xbf\xbf\xbf\xbf\xbf\xc3\x96z", -1, 1},
143 {ERE, "x.+z", "\xe2\x80\x94xz", -1, 1},
144 {ERE, "x.+z", "\xe2\x80\x94xyz", 3, 1},
145 {ERE, "x.+z", "\xe2\x80\x94x\xc3\x84y\xe2\x80\x94z", 3, 1},
146 {ERE, "x.+z", "\xe2\x80\x94x\xe2\x80\x94z", 3, 1},
147 {ERE, "x.+z", "\xe2\x80\x94x\xf0\x9d\x80\x80\xc3\x84z", 3, 1},
148 {ERE, "x.+z", "\xe2\x80\x94x.~\xe2\x80\x94\xf9\x81\x82\x83\x84z", 3, 1},
149 {ERE, "x.+z", "\xe2\x80\x94x\xfd\xbf\xbf\xbf\xbf\xbfz", 3, 1},
150 {ERE, "x.{1,2}z", "\xe2\x80\x94xz", -1, 1},
151 {ERE, "x.{1,2}z", "\xe2\x80\x94x\xc3\x96y\xc3\xa4z", -1, 1},
152 {ERE, "x.{1,2}z", "\xe2\x80\x94xyz", 3, 1},
153 {ERE, "x.{1,2}z", "\xe2\x80\x94x\xc3\x84\xe2\x80\x94z", 3, 1},
154 {ERE, "x.{1,2}z", "\xe2\x80\x94x\xe2\x80\x94z", 3, 1},
155 {ERE, "x.{1,2}z", "\xe2\x80\x94x\xf0\x9d\x80\x80\xc3\x84z", 3, 1},
156 {ERE, "x.{1,2}z", "\xe2\x80\x94x~\xe2\x80\x94z", 3, 1},
157 {ERE, "x.{1,2}z", "\xe2\x80\x94x\xfd\xbf\xbf\xbf\xbf\xbfz", 3, 1},
158 {ERE, "x(.w|\xc3\x86)?z", "axz", 1, 1},
159 {ERE, "x(.w|\xc3\x86)?z", "ax\xfd\xbf\xbf\xbf\xbf\xbfwz", 1, 1},
160 {ERE, "x(.w|\xc3\x86)?z", "ax\xc3\x86z", 1, 1},
161 {ERE, "x(.w|\xc3\x86)?z", "ax\xe2\x80\x96wz", 1, 1},
162 /* Should not be optimized. */
163 {BRE, "x[\xc3\x84\xc3\xa4]y", "ax\xc3\xa4y", 1, 0},
164 {BRE, "x[A-Z,]y", "axCy", 1, 0},
165 {BRE, "x[^y]z", "ax\xe2\x80\x94z", 1, 0},
166 {BRE, "x[[:alnum:]]z", "ax\xc3\x96z", 1, 0},
167 {BRE, "x[[=A=]]z", "axAz", 1, 0},
168 {BRE, "x[[=\xc3\x84=]]z", "ax\xc3\x84z", 1, 0},
169 {BRE, "\\<g", "\xe2\x80\x94g", 3, 0},
170 {BRE, "\\bg\\b", "\xe2\x80\x94g", 3, 0},
171 {BRE, "\\Bg\\B", "\xc3\xa4g\xc3\xa4", 2, 0},
172 {BRE, "a\\wz", "a\xc3\x84z", 0, 0},
173 {BRE, "x\\Wz", "\xc3\x96x\xe2\x80\x94z", 2, 0},
174 {ERE, "x[\xc3\x84\xc3\xa4]y", "ax\xc3\xa4y", 1, 0},
175 {ERE, "x[A-Z,]y", "axCy", 1, 0},
176 {ERE, "x[^y]z", "ax\xe2\x80\x94z", 1, 0},
177 {ERE, "x[[:alnum:]]z", "ax\xc3\x96z", 1, 0},
178 {ERE, "x[[=A=]]z", "axAz", 1, 0},
179 {ERE, "x[[=\xc3\x84=]]z", "ax\xc3\x84z", 1, 0},
180 {ERE, "\\<g", "\xe2\x80\x94g", 3, 0},
181 {ERE, "\\bg\\b", "\xe2\x80\x94g", 3, 0},
182 {ERE, "\\Bg\\B", "\xc3\xa4g\xc3\xa4", 2, 0},
183 {ERE, "a\\wz", "a\xc3\x84z", 0, 0},
184 {ERE, "x\\Wz", "\xc3\x96x\xe2\x80\x94z", 2, 0},
185 };
186
187 int
188 main (void)
189 {
190 struct re_pattern_buffer regbuf;
191 const char *err;
192 size_t i;
193 int ret = 0;
194
195 mtrace ();
196
197 setlocale (LC_ALL, "de_DE.UTF-8");
198 for (i = 0; i < sizeof (tests) / sizeof (tests[0]); ++i)
199 {
200 int res, optimized;
201
202 re_set_syntax (tests[i].syntax);
203 memset (®buf, '\0', sizeof (regbuf));
204 err = re_compile_pattern (tests[i].pattern, strlen (tests[i].pattern),
205 ®buf);
206 if (err != NULL)
207 {
208 printf ("re_compile_pattern failed: %s\n", err);
209 ret = 1;
210 continue;
211 }
212
213 /* Check if re_search will be done as multi-byte or single-byte. */
214 optimized = ((re_dfa_t *) regbuf.buffer)->mb_cur_max == 1;
215 if (optimized != tests[i].optimize)
216 {
217 printf ("pattern %zd %soptimized while it should%s be\n",
218 i, optimized ? "" : "not ", tests[i].optimize ? "" : " not");
219 ret = 1;
220 }
221
222 int str_len = strlen (tests[i].string);
223 res = re_search (®buf, tests[i].string, str_len, 0, str_len, NULL);
224 if (res != tests[i].res)
225 {
226 printf ("re_search %zd failed: %d\n", i, res);
227 ret = 1;
228 regfree (®buf);
229 continue;
230 }
231
232 res = re_search (®buf, tests[i].string, str_len, str_len, -str_len,
233 NULL);
234 if (res != tests[i].res)
235 {
236 printf ("backward re_search %zd failed: %d\n", i, res);
237 ret = 1;
238 regfree (®buf);
239 continue;
240 }
241 regfree (®buf);
242
243 re_set_syntax (tests[i].syntax | RE_ICASE);
244 memset (®buf, '\0', sizeof (regbuf));
245 err = re_compile_pattern (tests[i].pattern, strlen (tests[i].pattern),
246 ®buf);
247 if (err != NULL)
248 {
249 printf ("re_compile_pattern failed: %s\n", err);
250 ret = 1;
251 continue;
252 }
253
254 /* Check if re_search will be done as multi-byte or single-byte. */
255 optimized = ((re_dfa_t *) regbuf.buffer)->mb_cur_max == 1;
256 if (optimized)
257 {
258 printf ("pattern %zd optimized while it should not be when case insensitive\n",
259 i);
260 ret = 1;
261 }
262
263 res = re_search (®buf, tests[i].string, str_len, 0, str_len, NULL);
264 if (res != tests[i].res)
265 {
266 printf ("ICASE re_search %zd failed: %d\n", i, res);
267 ret = 1;
268 regfree (®buf);
269 continue;
270 }
271
272 res = re_search (®buf, tests[i].string, str_len, str_len, -str_len,
273 NULL);
274 if (res != tests[i].res)
275 {
276 printf ("ICASE backward re_search %zd failed: %d\n", i, res);
277 ret = 1;
278 regfree (®buf);
279 continue;
280 }
281 regfree (®buf);
282 }
283
284 return ret;
285 }