1 # serial 53
3 # Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2003, 2004, 2005,
4 # 2006, 2007, 2008, 2009 Free Software Foundation, Inc.
5 #
6 # This file is free software; the Free Software Foundation
7 # gives unlimited permission to copy and/or distribute it,
8 # with or without modifications, as long as this notice is preserved.
10 dnl Initially derived from code in GNU grep.
11 dnl Mostly written by Jim Meyering.
13 AC_PREREQ([2.50])
15 AC_DEFUN([gl_REGEX],
16 [
17 AC_CHECK_HEADERS_ONCE([locale.h])
19 AC_ARG_WITH([included-regex],
20 [AS_HELP_STRING([--without-included-regex],
21 [don't compile regex; this is the default on 32-bit
22 systems with recent-enough versions of the GNU C
23 Library (use with caution on other systems).
24 On systems with 64-bit ptrdiff_t and 32-bit int,
25 --with-included-regex is the default, in case
26 regex functions operate on very long strings (>2GB)])])
28 case $with_included_regex in #(
29 yes|no) ac_use_included_regex=$with_included_regex
30 ;;
31 '')
32 # If the system regex support is good enough that it passes the
33 # following run test, then default to *not* using the included regex.c.
34 # If cross compiling, assume the test would fail and use the included
35 # regex.c.
36 AC_CACHE_CHECK([for working re_compile_pattern],
37 [gl_cv_func_re_compile_pattern_working],
38 [AC_RUN_IFELSE(
39 [AC_LANG_PROGRAM(
40 [AC_INCLUDES_DEFAULT[
41 #if HAVE_LOCALE_H
42 #include <locale.h>
43 #endif
44 #include <limits.h>
45 #include <regex.h>
46 ]],
47 [[static struct re_pattern_buffer regex;
48 unsigned char folded_chars[UCHAR_MAX + 1];
49 int i;
50 const char *s;
51 struct re_registers regs;
53 #if HAVE_LOCALE_H
54 /* http://sourceware.org/ml/libc-hacker/2006-09/msg00008.html
55 This test needs valgrind to catch the bug on Debian
56 GNU/Linux 3.1 x86, but it might catch the bug better
57 on other platforms and it shouldn't hurt to try the
58 test here. */
59 if (setlocale (LC_ALL, "en_US.UTF-8"))
60 {
61 static char const pat[] = "insert into";
62 static char const data[] =
63 "\xFF\0\x12\xA2\xAA\xC4\xB1,K\x12\xC4\xB1*\xACK";
64 re_set_syntax (RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE
65 | RE_ICASE);
66 memset (®ex, 0, sizeof regex);
67 s = re_compile_pattern (pat, sizeof pat - 1, ®ex);
68 if (s)
69 return 1;
70 if (re_search (®ex, data, sizeof data - 1,
71 0, sizeof data - 1, ®s)
72 != -1)
73 return 1;
74 if (! setlocale (LC_ALL, "C"))
75 return 1;
76 }
77 #endif
79 /* This test is from glibc bug 3957, reported by Andrew Mackey. */
80 re_set_syntax (RE_SYNTAX_EGREP | RE_HAT_LISTS_NOT_NEWLINE);
81 memset (®ex, 0, sizeof regex);
82 s = re_compile_pattern ("a[^x]b", 6, ®ex);
83 if (s)
84 return 1;
86 /* This should fail, but succeeds for glibc-2.5. */
87 if (re_search (®ex, "a\nb", 3, 0, 3, ®s) != -1)
88 return 1;
90 /* This regular expression is from Spencer ere test number 75
91 in grep-2.3. */
92 re_set_syntax (RE_SYNTAX_POSIX_EGREP);
93 memset (®ex, 0, sizeof regex);
94 for (i = 0; i <= UCHAR_MAX; i++)
95 folded_chars[i] = i;
96 regex.translate = folded_chars;
97 s = re_compile_pattern ("a[[:@:>@:]]b\n", 11, ®ex);
98 /* This should fail with _Invalid character class name_ error. */
99 if (!s)
100 return 1;
102 /* This should succeed, but does not for glibc-2.1.3. */
103 memset (®ex, 0, sizeof regex);
104 s = re_compile_pattern ("{1", 2, ®ex);
106 if (s)
107 return 1;
109 /* The following example is derived from a problem report
110 against gawk from Jorge Stolfi <stolfi@ic.unicamp.br>. */
111 memset (®ex, 0, sizeof regex);
112 s = re_compile_pattern ("[an\371]*n", 7, ®ex);
113 if (s)
114 return 1;
116 /* This should match, but does not for glibc-2.2.1. */
117 if (re_match (®ex, "an", 2, 0, ®s) != 2)
118 return 1;
120 memset (®ex, 0, sizeof regex);
121 s = re_compile_pattern ("x", 1, ®ex);
122 if (s)
123 return 1;
125 /* glibc-2.2.93 does not work with a negative RANGE argument. */
126 if (re_search (®ex, "wxy", 3, 2, -2, ®s) != 1)
127 return 1;
129 /* The version of regex.c in older versions of gnulib
130 ignored RE_ICASE. Detect that problem too. */
131 re_set_syntax (RE_SYNTAX_EMACS | RE_ICASE);
132 memset (®ex, 0, sizeof regex);
133 s = re_compile_pattern ("x", 1, ®ex);
134 if (s)
135 return 1;
137 if (re_search (®ex, "WXY", 3, 0, 3, ®s) < 0)
138 return 1;
140 /* Catch a bug reported by Vin Shelton in
141 http://lists.gnu.org/archive/html/bug-coreutils/2007-06/msg00089.html
142 */
143 re_set_syntax (RE_SYNTAX_POSIX_BASIC
144 & ~RE_CONTEXT_INVALID_DUP
145 & ~RE_NO_EMPTY_RANGES);
146 memset (®ex, 0, sizeof regex);
147 s = re_compile_pattern ("[[:alnum:]_-]\\\\+$", 16, ®ex);
148 if (s)
149 return 1;
151 /* REG_STARTEND was added to glibc on 2004-01-15.
152 Reject older versions. */
153 if (! REG_STARTEND)
154 return 1;
156 /* Reject hosts whose regoff_t values are too narrow.
157 These include glibc 2.3.5 on hosts with 64-bit ptrdiff_t
158 and 32-bit int. */
159 if (sizeof (regoff_t) < sizeof (ptrdiff_t)
160 || sizeof (regoff_t) < sizeof (ssize_t))
161 return 1;
163 return 0;]])],
164 [gl_cv_func_re_compile_pattern_working=yes],
165 [gl_cv_func_re_compile_pattern_working=no],
166 dnl When crosscompiling, assume it is not working.
167 [gl_cv_func_re_compile_pattern_working=no])])
168 case $gl_cv_func_re_compile_pattern_working in #(
169 yes) ac_use_included_regex=no;; #(
170 no) ac_use_included_regex=yes;;
171 esac
172 ;;
173 *) AC_MSG_ERROR([Invalid value for --with-included-regex: $with_included_regex])
174 ;;
175 esac
177 if test $ac_use_included_regex = yes; then
178 AC_DEFINE([_REGEX_LARGE_OFFSETS], [1],
179 [Define if you want regoff_t to be at least as wide POSIX requires.])
180 AC_DEFINE([re_syntax_options], [rpl_re_syntax_options],
181 [Define to rpl_re_syntax_options if the replacement should be used.])
182 AC_DEFINE([re_set_syntax], [rpl_re_set_syntax],
183 [Define to rpl_re_set_syntax if the replacement should be used.])
184 AC_DEFINE([re_compile_pattern], [rpl_re_compile_pattern],
185 [Define to rpl_re_compile_pattern if the replacement should be used.])
186 AC_DEFINE([re_compile_fastmap], [rpl_re_compile_fastmap],
187 [Define to rpl_re_compile_fastmap if the replacement should be used.])
188 AC_DEFINE([re_search], [rpl_re_search],
189 [Define to rpl_re_search if the replacement should be used.])
190 AC_DEFINE([re_search_2], [rpl_re_search_2],
191 [Define to rpl_re_search_2 if the replacement should be used.])
192 AC_DEFINE([re_match], [rpl_re_match],
193 [Define to rpl_re_match if the replacement should be used.])
194 AC_DEFINE([re_match_2], [rpl_re_match_2],
195 [Define to rpl_re_match_2 if the replacement should be used.])
196 AC_DEFINE([re_set_registers], [rpl_re_set_registers],
197 [Define to rpl_re_set_registers if the replacement should be used.])
198 AC_DEFINE([re_comp], [rpl_re_comp],
199 [Define to rpl_re_comp if the replacement should be used.])
200 AC_DEFINE([re_exec], [rpl_re_exec],
201 [Define to rpl_re_exec if the replacement should be used.])
202 AC_DEFINE([regcomp], [rpl_regcomp],
203 [Define to rpl_regcomp if the replacement should be used.])
204 AC_DEFINE([regexec], [rpl_regexec],
205 [Define to rpl_regexec if the replacement should be used.])
206 AC_DEFINE([regerror], [rpl_regerror],
207 [Define to rpl_regerror if the replacement should be used.])
208 AC_DEFINE([regfree], [rpl_regfree],
209 [Define to rpl_regfree if the replacement should be used.])
210 AC_LIBOBJ([regex])
211 gl_PREREQ_REGEX
212 fi
213 ])
215 # Prerequisites of lib/regex.c and lib/regex_internal.c.
216 AC_DEFUN([gl_PREREQ_REGEX],
217 [
218 AC_REQUIRE([AC_USE_SYSTEM_EXTENSIONS])
219 AC_REQUIRE([AC_C_RESTRICT])
220 AC_REQUIRE([AC_TYPE_MBSTATE_T])
221 AC_CHECK_HEADERS([libintl.h])
222 AC_CHECK_FUNCS_ONCE([isblank iswctype wcscoll])
223 AC_CHECK_DECLS([isblank], [], [], [#include <ctype.h>])
224 ])