3 # Copyright (C) 1996-2001, 2003-2020 Free Software Foundation, Inc.
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.
9 dnl Initially derived from code in GNU grep.
10 dnl Mostly written by Jim Meyering.
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).])])
23 case $with_included_regex in #(
24 yes|no) ac_use_included_regex=$with_included_regex
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
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],
43 #if defined M_CHECK_ACTION || HAVE_DECL_ALARM
53 /* Exit with distinguishable exit code. */
54 static void sigabrt_no_core (int sig) { raise (SIGTERM); }
58 static struct re_pattern_buffer regex;
59 unsigned char folded_chars[UCHAR_MAX + 1];
62 struct re_registers regs;
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. */
68 signal (SIGALRM, SIG_DFL);
72 signal (SIGABRT, sigabrt_no_core);
73 mallopt (M_CHECK_ACTION, 2);
76 if (setlocale (LC_ALL, "en_US.UTF-8"))
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
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
89 memset (®ex, 0, sizeof regex);
90 s = re_compile_pattern (pat, sizeof pat - 1, ®ex);
95 if (re_search (®ex, data, sizeof data - 1,
96 0, sizeof data - 1, ®s)
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>.
108 static char const pat[] = "[^x]x";
109 static char const data[] =
110 /* <U1000><U103B><U103D><U1014><U103A><U102F><U1015><U103A> */
121 memset (®ex, 0, sizeof regex);
122 s = re_compile_pattern (pat, sizeof pat - 1, ®ex);
127 i = re_search (®ex, data, sizeof data - 1,
128 0, sizeof data - 1, 0);
129 if (i != 0 && i != 21)
135 if (! setlocale (LC_ALL, "C"))
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 (®ex, 0, sizeof regex);
142 s = re_compile_pattern ("a[^x]b", 6, ®ex);
147 /* This should fail, but succeeds for glibc-2.5. */
148 if (re_search (®ex, "a\nb", 3, 0, 3, ®s) != -1)
153 /* This regular expression is from Spencer ere test number 75
155 re_set_syntax (RE_SYNTAX_POSIX_EGREP);
156 memset (®ex, 0, sizeof regex);
157 for (i = 0; i <= UCHAR_MAX; i++)
159 regex.translate = folded_chars;
160 s = re_compile_pattern ("a[[:@:>@:]]b\n", 11, ®ex);
161 /* This should fail with _Invalid character class name_ error. */
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 (®ex, 0, sizeof regex);
172 s = re_compile_pattern ("a[b-a]", 6, ®ex);
179 /* This should succeed, but does not for glibc-2.1.3. */
180 memset (®ex, 0, sizeof regex);
181 s = re_compile_pattern ("{1", 2, ®ex);
187 /* The following example is derived from a problem report
188 against gawk from Jorge Stolfi <stolfi@ic.unicamp.br>. */
189 memset (®ex, 0, sizeof regex);
190 s = re_compile_pattern ("[an\371]*n", 7, ®ex);
195 /* This should match, but does not for glibc-2.2.1. */
196 if (re_match (®ex, "an", 2, 0, ®s) != 2)
206 memset (®ex, 0, sizeof regex);
207 s = re_compile_pattern ("x", 1, ®ex);
212 /* glibc-2.2.93 does not work with a negative RANGE argument. */
213 if (re_search (®ex, "wxy", 3, 2, -2, ®s) != 1)
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 (®ex, 0, sizeof regex);
227 s = re_compile_pattern ("x", 1, ®ex);
232 if (re_search (®ex, "WXY", 3, 0, 3, ®s) < 0)
242 /* Catch a bug reported by Vin Shelton in
243 https://lists.gnu.org/r/bug-coreutils/2007-06/msg00089.html
245 re_set_syntax (RE_SYNTAX_POSIX_BASIC
246 & ~RE_CONTEXT_INVALID_DUP
247 & ~RE_NO_EMPTY_RANGES);
248 memset (®ex, 0, sizeof regex);
249 s = re_compile_pattern ("[[:alnum:]_-]\\\\+$", 16, ®ex);
255 /* REG_STARTEND was added to glibc on 2004-01-15.
256 Reject older versions. */
260 /* Matching with the compiled form of this regexp would provoke
261 an assertion failure prior to glibc-2.28:
262 regexec.c:1375: pop_fail_stack: Assertion 'num >= 0' failed
263 With glibc-2.28, compilation fails and reports the invalid
265 re_set_syntax (RE_SYNTAX_POSIX_EGREP);
266 memset (®ex, 0, sizeof regex);
267 s = re_compile_pattern ("0|()0|\\1|0", 10, ®ex);
272 if (strcmp (s, "Invalid back reference"))
278 /* It would be nice to reject hosts whose regoff_t values are too
279 narrow (including glibc on hosts with 64-bit ptrdiff_t and
280 32-bit int), but we should wait until glibc implements this
281 feature. Otherwise, support for equivalence classes and
282 multibyte collation symbols would always be broken except
283 when compiling --without-included-regex. */
284 if (sizeof (regoff_t) < sizeof (ptrdiff_t)
285 || sizeof (regoff_t) < sizeof (ssize_t))
291 [gl_cv_func_re_compile_pattern_working=yes],
292 [gl_cv_func_re_compile_pattern_working=no],
294 # Guess no on native Windows.
295 mingw*) gl_cv_func_re_compile_pattern_working="guessing no" ;;
296 # Otherwise obey --enable-cross-guesses.
297 *) gl_cv_func_re_compile_pattern_working="$gl_cross_guess_normal" ;;
301 case "$gl_cv_func_re_compile_pattern_working" in #(
302 *yes) ac_use_included_regex=no;; #(
303 *no) ac_use_included_regex=yes;;
306 *) AC_MSG_ERROR([Invalid value for --with-included-regex: $with_included_regex])
310 if test $ac_use_included_regex = yes; then
311 AC_DEFINE([_REGEX_INCLUDE_LIMITS_H], [1],
312 [Define if you want <regex.h> to include <limits.h>, so that it
313 consistently overrides <limits.h>'s RE_DUP_MAX.])
314 AC_DEFINE([_REGEX_LARGE_OFFSETS], [1],
315 [Define if you want regoff_t to be at least as wide POSIX requires.])
316 AC_DEFINE([re_syntax_options], [rpl_re_syntax_options],
317 [Define to rpl_re_syntax_options if the replacement should be used.])
318 AC_DEFINE([re_set_syntax], [rpl_re_set_syntax],
319 [Define to rpl_re_set_syntax if the replacement should be used.])
320 AC_DEFINE([re_compile_pattern], [rpl_re_compile_pattern],
321 [Define to rpl_re_compile_pattern if the replacement should be used.])
322 AC_DEFINE([re_compile_fastmap], [rpl_re_compile_fastmap],
323 [Define to rpl_re_compile_fastmap if the replacement should be used.])
324 AC_DEFINE([re_search], [rpl_re_search],
325 [Define to rpl_re_search if the replacement should be used.])
326 AC_DEFINE([re_search_2], [rpl_re_search_2],
327 [Define to rpl_re_search_2 if the replacement should be used.])
328 AC_DEFINE([re_match], [rpl_re_match],
329 [Define to rpl_re_match if the replacement should be used.])
330 AC_DEFINE([re_match_2], [rpl_re_match_2],
331 [Define to rpl_re_match_2 if the replacement should be used.])
332 AC_DEFINE([re_set_registers], [rpl_re_set_registers],
333 [Define to rpl_re_set_registers if the replacement should be used.])
334 AC_DEFINE([re_comp], [rpl_re_comp],
335 [Define to rpl_re_comp if the replacement should be used.])
336 AC_DEFINE([re_exec], [rpl_re_exec],
337 [Define to rpl_re_exec if the replacement should be used.])
338 AC_DEFINE([regcomp], [rpl_regcomp],
339 [Define to rpl_regcomp if the replacement should be used.])
340 AC_DEFINE([regexec], [rpl_regexec],
341 [Define to rpl_regexec if the replacement should be used.])
342 AC_DEFINE([regerror], [rpl_regerror],
343 [Define to rpl_regerror if the replacement should be used.])
344 AC_DEFINE([regfree], [rpl_regfree],
345 [Define to rpl_regfree if the replacement should be used.])
349 # Prerequisites of lib/regex.c and lib/regex_internal.c.
350 AC_DEFUN([gl_PREREQ_REGEX],
352 AC_REQUIRE([AC_USE_SYSTEM_EXTENSIONS])
353 AC_REQUIRE([AC_C_INLINE])
354 AC_REQUIRE([AC_C_RESTRICT])
355 AC_REQUIRE([AC_TYPE_MBSTATE_T])
356 AC_REQUIRE([gl_EEMALLOC])
357 AC_REQUIRE([gl_GLIBC21])
358 AC_CHECK_HEADERS([libintl.h])
359 AC_CHECK_FUNCS_ONCE([isblank iswctype])
360 AC_CHECK_DECLS([isblank], [], [], [[#include <ctype.h>]])