source: trunk/src/sed/m4/regex.m4

Last change on this file was 3611, checked in by bird, 14 months ago

vendor/sed/current: GNU sed 4.9 (sed-4.9.tar.xz sha256:6e226b732e1cd739464ad6862bd1a1aba42d7982922da7a53519631d24975181)
File size: 15.0 KB
Line 
1# serial 73
2
3# Copyright (C) 1996-2001, 2003-2022 Free Software Foundation, Inc.
4#
5# This file is free software; the Free Software Foundation
6# gives unlimited permission to copy and/or distribute it,
7# with or without modifications, as long as this notice is preserved.
8
9dnl Initially derived from code in GNU grep.
10dnl Mostly written by Jim Meyering.
11
12AC_PREREQ([2.50])
13
14AC_DEFUN([gl_REGEX],
15[
16 AC_REQUIRE([AC_CANONICAL_HOST]) dnl for cross-compiles
17 AC_ARG_WITH([included-regex],
18 [AS_HELP_STRING([--without-included-regex],
19 [don't compile regex; this is the default on systems
20 with recent-enough versions of the GNU C Library
21 (use with caution on other systems).])])
22
23 case $with_included_regex in #(
24 yes|no) ac_use_included_regex=$with_included_regex
25 ;;
26 '')
27 # If the system regex support is good enough that it passes the
28 # following run test, then default to *not* using the included regex.c.
29 # If cross compiling, assume the test would fail and use the included
30 # regex.c.
31 AC_CHECK_DECLS_ONCE([alarm])
32 AC_CHECK_HEADERS_ONCE([malloc.h])
33 AC_CACHE_CHECK([for working re_compile_pattern],
34 [gl_cv_func_re_compile_pattern_working],
35 [AC_RUN_IFELSE(
36 [AC_LANG_PROGRAM(
37 [[#include <regex.h>
38
39 #include <locale.h>
40 #include <limits.h>
41 #include <string.h>
42
43 #if defined M_CHECK_ACTION || HAVE_DECL_ALARM
44 # include <signal.h>
45 # include <unistd.h>
46 #endif
47
48 #if HAVE_MALLOC_H
49 # include <malloc.h>
50 #endif
51
52 #ifdef M_CHECK_ACTION
53 /* Exit with distinguishable exit code. */
54 static void sigabrt_no_core (int sig) { raise (SIGTERM); }
55 #endif
56 ]],
57 [[int result = 0;
58 static struct re_pattern_buffer regex;
59 unsigned char folded_chars[UCHAR_MAX + 1];
60 int i;
61 const char *s;
62 struct re_registers regs;
63
64 /* Some builds of glibc go into an infinite loop on this
65 test. Use alarm to force death, and mallopt to avoid
66 malloc recursion in diagnosing the corrupted heap. */
67#if HAVE_DECL_ALARM
68 signal (SIGALRM, SIG_DFL);
69 alarm (2);
70#endif
71#ifdef M_CHECK_ACTION
72 signal (SIGABRT, sigabrt_no_core);
73 mallopt (M_CHECK_ACTION, 2);
74#endif
75
76 if (setlocale (LC_ALL, "en_US.UTF-8"))
77 {
78 {
79 /* https://sourceware.org/ml/libc-hacker/2006-09/msg00008.html
80 This test needs valgrind to catch the bug on Debian
81 GNU/Linux 3.1 x86, but it might catch the bug better
82 on other platforms and it shouldn't hurt to try the
83 test here. */
84 static char const pat[] = "insert into";
85 static char const data[] =
86 "\xFF\0\x12\xA2\xAA\xC4\xB1,K\x12\xC4\xB1*\xACK";
87 re_set_syntax (RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE
88 | RE_ICASE);
89 memset (&regex, 0, sizeof regex);
90 s = re_compile_pattern (pat, sizeof pat - 1, &regex);
91 if (s)
92 result |= 1;
93 else
94 {
95 if (re_search (&regex, data, sizeof data - 1,
96 0, sizeof data - 1, &regs)
97 != -1)
98 result |= 1;
99 regfree (&regex);
100 }
101 }
102
103 {
104 /* This test is from glibc bug 15078.
105 The test case is from Andreas Schwab in
106 <https://sourceware.org/ml/libc-alpha/2013-01/msg00967.html>.
107 */
108 static char const pat[] = "[^x]x";
109 static char const data[] =
110 /* <U1000><U103B><U103D><U1014><U103A><U102F><U1015><U103A> */
111 "\xe1\x80\x80"
112 "\xe1\x80\xbb"
113 "\xe1\x80\xbd"
114 "\xe1\x80\x94"
115 "\xe1\x80\xba"
116 "\xe1\x80\xaf"
117 "\xe1\x80\x95"
118 "\xe1\x80\xba"
119 "x";
120 re_set_syntax (0);
121 memset (&regex, 0, sizeof regex);
122 s = re_compile_pattern (pat, sizeof pat - 1, &regex);
123 if (s)
124 result |= 1;
125 else
126 {
127 i = re_search (&regex, data, sizeof data - 1,
128 0, sizeof data - 1, 0);
129 if (i != 0 && i != 21)
130 result |= 1;
131 regfree (&regex);
132 }
133 }
134
135 if (! setlocale (LC_ALL, "C"))
136 return 1;
137 }
138
139 /* This test is from glibc bug 3957, reported by Andrew Mackey. */
140 re_set_syntax (RE_SYNTAX_EGREP | RE_HAT_LISTS_NOT_NEWLINE);
141 memset (&regex, 0, sizeof regex);
142 s = re_compile_pattern ("a[^x]b", 6, &regex);
143 if (s)
144 result |= 2;
145 else
146 {
147 /* This should fail, but succeeds for glibc-2.5. */
148 if (re_search (&regex, "a\nb", 3, 0, 3, &regs) != -1)
149 result |= 2;
150 regfree (&regex);
151 }
152
153 /* This regular expression is from Spencer ere test number 75
154 in grep-2.3. */
155 re_set_syntax (RE_SYNTAX_POSIX_EGREP);
156 memset (&regex, 0, sizeof regex);
157 for (i = 0; i <= UCHAR_MAX; i++)
158 folded_chars[i] = i;
159 regex.translate = folded_chars;
160 s = re_compile_pattern ("a[[:@:>@:]]b\n", 11, &regex);
161 /* This should fail with _Invalid character class name_ error. */
162 if (!s)
163 {
164 result |= 4;
165 regfree (&regex);
166 }
167
168 /* Ensure that [b-a] is diagnosed as invalid, when
169 using RE_NO_EMPTY_RANGES. */
170 re_set_syntax (RE_SYNTAX_POSIX_EGREP | RE_NO_EMPTY_RANGES);
171 memset (&regex, 0, sizeof regex);
172 s = re_compile_pattern ("a[b-a]", 6, &regex);
173 if (s == 0)
174 {
175 result |= 8;
176 regfree (&regex);
177 }
178
179 /* This should succeed, but does not for glibc-2.1.3. */
180 memset (&regex, 0, sizeof regex);
181 s = re_compile_pattern ("{1", 2, &regex);
182 if (s)
183 result |= 8;
184 else
185 regfree (&regex);
186
187 /* The following example is derived from a problem report
188 against gawk from Jorge Stolfi <stolfi@ic.unicamp.br>. */
189 memset (&regex, 0, sizeof regex);
190 s = re_compile_pattern ("[an\371]*n", 7, &regex);
191 if (s)
192 result |= 8;
193 else
194 {
195 /* This should match, but does not for glibc-2.2.1. */
196 if (re_match (&regex, "an", 2, 0, &regs) != 2)
197 result |= 8;
198 else
199 {
200 free (regs.start);
201 free (regs.end);
202 }
203 regfree (&regex);
204 }
205
206 memset (&regex, 0, sizeof regex);
207 s = re_compile_pattern ("x", 1, &regex);
208 if (s)
209 result |= 8;
210 else
211 {
212 /* glibc-2.2.93 does not work with a negative RANGE argument. */
213 if (re_search (&regex, "wxy", 3, 2, -2, &regs) != 1)
214 result |= 8;
215 else
216 {
217 free (regs.start);
218 free (regs.end);
219 }
220 regfree (&regex);
221 }
222
223 /* The version of regex.c in older versions of gnulib
224 ignored RE_ICASE. Detect that problem too. */
225 re_set_syntax (RE_SYNTAX_EMACS | RE_ICASE);
226 memset (&regex, 0, sizeof regex);
227 s = re_compile_pattern ("x", 1, &regex);
228 if (s)
229 result |= 16;
230 else
231 {
232 if (re_search (&regex, "WXY", 3, 0, 3, &regs) < 0)
233 result |= 16;
234 else
235 {
236 free (regs.start);
237 free (regs.end);
238 }
239 regfree (&regex);
240 }
241
242 /* Catch a bug reported by Vin Shelton in
243 https://lists.gnu.org/r/bug-coreutils/2007-06/msg00089.html
244 */
245 re_set_syntax (RE_SYNTAX_POSIX_BASIC
246 & ~RE_CONTEXT_INVALID_DUP
247 & ~RE_NO_EMPTY_RANGES);
248 memset (&regex, 0, sizeof regex);
249 s = re_compile_pattern ("[[:alnum:]_-]\\\\+\$", 16, &regex);
250 if (s)
251 result |= 32;
252 else
253 regfree (&regex);
254
255 /* REG_STARTEND was added to glibc on 2004-01-15.
256 Reject older versions. */
257 if (! REG_STARTEND)
258 result |= 64;
259
260 /* Matching with the compiled form of this regexp would provoke
261 an assertion failure prior to glibc-2.28:
262 regexec.c:1375: pop_fail_stack: Assertion 'num >= 0' failed
263 With glibc-2.28, compilation fails and reports the invalid
264 back reference. */
265 re_set_syntax (RE_SYNTAX_POSIX_EGREP);
266 memset (&regex, 0, sizeof regex);
267 s = re_compile_pattern ("0|()0|\\\\1|0", 10, &regex);
268 if (!s)
269 {
270 memset (&regs, 0, sizeof regs);
271 i = re_search (&regex, "x", 1, 0, 1, &regs);
272 if (i != -1)
273 result |= 64;
274 if (0 <= i)
275 {
276 free (regs.start);
277 free (regs.end);
278 }
279 regfree (&regex);
280 }
281 else
282 {
283 if (strcmp (s, "Invalid back reference"))
284 result |= 64;
285 }
286
287 /* glibc bug 11053. */
288 re_set_syntax (RE_SYNTAX_POSIX_BASIC);
289 memset (&regex, 0, sizeof regex);
290 static char const pat_sub2[] = "\\\\(a*\\\\)*a*\\\\1";
291 s = re_compile_pattern (pat_sub2, sizeof pat_sub2 - 1, &regex);
292 if (s)
293 result |= 64;
294 else
295 {
296 memset (&regs, 0, sizeof regs);
297 static char const data[] = "a";
298 int datalen = sizeof data - 1;
299 i = re_search (&regex, data, datalen, 0, datalen, &regs);
300 if (i != 0)
301 result |= 64;
302 else if (regs.num_regs < 2)
303 result |= 64;
304 else if (! (regs.start[0] == 0 && regs.end[0] == 1))
305 result |= 64;
306 else if (! (regs.start[1] == 0 && regs.end[1] == 0))
307 result |= 64;
308 regfree (&regex);
309 free (regs.start);
310 free (regs.end);
311 }
312
313#if 0
314 /* It would be nice to reject hosts whose regoff_t values are too
315 narrow (including glibc on hosts with 64-bit ptrdiff_t and
316 32-bit int), but we should wait until glibc implements this
317 feature. Otherwise, support for equivalence classes and
318 multibyte collation symbols would always be broken except
319 when compiling --without-included-regex. */
320 if (sizeof (regoff_t) < sizeof (ptrdiff_t)
321 || sizeof (regoff_t) < sizeof (ssize_t))
322 result |= 64;
323#endif
324
325 return result;
326 ]])],
327 [gl_cv_func_re_compile_pattern_working=yes],
328 [gl_cv_func_re_compile_pattern_working=no],
329 [case "$host_os" in
330 # Guess no on native Windows.
331 mingw*) gl_cv_func_re_compile_pattern_working="guessing no" ;;
332 # Otherwise obey --enable-cross-guesses.
333 *) gl_cv_func_re_compile_pattern_working="$gl_cross_guess_normal" ;;
334 esac
335 ])
336 ])
337 case "$gl_cv_func_re_compile_pattern_working" in #(
338 *yes) ac_use_included_regex=no;; #(
339 *no) ac_use_included_regex=yes;;
340 esac
341 ;;
342 *) AC_MSG_ERROR([Invalid value for --with-included-regex: $with_included_regex])
343 ;;
344 esac
345
346 if test $ac_use_included_regex = yes; then
347 AC_DEFINE([_REGEX_INCLUDE_LIMITS_H], [1],
348 [Define if you want <regex.h> to include <limits.h>, so that it
349 consistently overrides <limits.h>'s RE_DUP_MAX.])
350 AC_DEFINE([_REGEX_LARGE_OFFSETS], [1],
351 [Define if you want regoff_t to be at least as wide POSIX requires.])
352 AC_DEFINE([re_syntax_options], [rpl_re_syntax_options],
353 [Define to rpl_re_syntax_options if the replacement should be used.])
354 AC_DEFINE([re_set_syntax], [rpl_re_set_syntax],
355 [Define to rpl_re_set_syntax if the replacement should be used.])
356 AC_DEFINE([re_compile_pattern], [rpl_re_compile_pattern],
357 [Define to rpl_re_compile_pattern if the replacement should be used.])
358 AC_DEFINE([re_compile_fastmap], [rpl_re_compile_fastmap],
359 [Define to rpl_re_compile_fastmap if the replacement should be used.])
360 AC_DEFINE([re_search], [rpl_re_search],
361 [Define to rpl_re_search if the replacement should be used.])
362 AC_DEFINE([re_search_2], [rpl_re_search_2],
363 [Define to rpl_re_search_2 if the replacement should be used.])
364 AC_DEFINE([re_match], [rpl_re_match],
365 [Define to rpl_re_match if the replacement should be used.])
366 AC_DEFINE([re_match_2], [rpl_re_match_2],
367 [Define to rpl_re_match_2 if the replacement should be used.])
368 AC_DEFINE([re_set_registers], [rpl_re_set_registers],
369 [Define to rpl_re_set_registers if the replacement should be used.])
370 AC_DEFINE([re_comp], [rpl_re_comp],
371 [Define to rpl_re_comp if the replacement should be used.])
372 AC_DEFINE([re_exec], [rpl_re_exec],
373 [Define to rpl_re_exec if the replacement should be used.])
374 AC_DEFINE([regcomp], [rpl_regcomp],
375 [Define to rpl_regcomp if the replacement should be used.])
376 AC_DEFINE([regexec], [rpl_regexec],
377 [Define to rpl_regexec if the replacement should be used.])
378 AC_DEFINE([regerror], [rpl_regerror],
379 [Define to rpl_regerror if the replacement should be used.])
380 AC_DEFINE([regfree], [rpl_regfree],
381 [Define to rpl_regfree if the replacement should be used.])
382 fi
383])
384
385# Prerequisites of lib/regex.c and lib/regex_internal.c.
386AC_DEFUN([gl_PREREQ_REGEX],
387[
388 AC_REQUIRE([AC_USE_SYSTEM_EXTENSIONS])
389 AC_REQUIRE([AC_C_INLINE])
390 AC_REQUIRE([AC_C_RESTRICT])
391 AC_REQUIRE([AC_TYPE_MBSTATE_T])
392 AC_REQUIRE([gl_EEMALLOC])
393 AC_CHECK_HEADERS([libintl.h])
394 AC_CHECK_FUNCS_ONCE([isblank iswctype])
395 AC_CHECK_DECLS([isblank], [], [], [[#include <ctype.h>]])
396])
Note: See TracBrowser for help on using the repository browser.