1 # serial 54
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 systems
22 with recent-enough versions of the GNU C Library
23 (use with caution on other systems).])])
25 case $with_included_regex in #(
26 yes|no) ac_use_included_regex=$with_included_regex
27 ;;
28 '')
29 # If the system regex support is good enough that it passes the
30 # following run test, then default to *not* using the included regex.c.
31 # If cross compiling, assume the test would fail and use the included
32 # regex.c.
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 [AC_INCLUDES_DEFAULT[
38 #if HAVE_LOCALE_H
39 #include <locale.h>
40 #endif
41 #include <limits.h>
42 #include <regex.h>
43 ]],
44 [[static struct re_pattern_buffer regex;
45 unsigned char folded_chars[UCHAR_MAX + 1];
46 int i;
47 const char *s;
48 struct re_registers regs;
50 #if HAVE_LOCALE_H
51 /* http://sourceware.org/ml/libc-hacker/2006-09/msg00008.html
52 This test needs valgrind to catch the bug on Debian
53 GNU/Linux 3.1 x86, but it might catch the bug better
54 on other platforms and it shouldn't hurt to try the
55 test here. */
56 if (setlocale (LC_ALL, "en_US.UTF-8"))
57 {
58 static char const pat[] = "insert into";
59 static char const data[] =
60 "\xFF\0\x12\xA2\xAA\xC4\xB1,K\x12\xC4\xB1*\xACK";
61 re_set_syntax (RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE
62 | RE_ICASE);
63 memset (®ex, 0, sizeof regex);
64 s = re_compile_pattern (pat, sizeof pat - 1, ®ex);
65 if (s)
66 return 1;
67 if (re_search (®ex, data, sizeof data - 1,
68 0, sizeof data - 1, ®s)
69 != -1)
70 return 1;
71 if (! setlocale (LC_ALL, "C"))
72 return 1;
73 }
74 #endif
76 /* This test is from glibc bug 3957, reported by Andrew Mackey. */
77 re_set_syntax (RE_SYNTAX_EGREP | RE_HAT_LISTS_NOT_NEWLINE);
78 memset (®ex, 0, sizeof regex);
79 s = re_compile_pattern ("a[^x]b", 6, ®ex);
80 if (s)
81 return 1;
83 /* This should fail, but succeeds for glibc-2.5. */
84 if (re_search (®ex, "a\nb", 3, 0, 3, ®s) != -1)
85 return 1;
87 /* This regular expression is from Spencer ere test number 75
88 in grep-2.3. */
89 re_set_syntax (RE_SYNTAX_POSIX_EGREP);
90 memset (®ex, 0, sizeof regex);
91 for (i = 0; i <= UCHAR_MAX; i++)
92 folded_chars[i] = i;
93 regex.translate = folded_chars;
94 s = re_compile_pattern ("a[[:@:>@:]]b\n", 11, ®ex);
95 /* This should fail with _Invalid character class name_ error. */
96 if (!s)
97 return 1;
99 /* This should succeed, but does not for glibc-2.1.3. */
100 memset (®ex, 0, sizeof regex);
101 s = re_compile_pattern ("{1", 2, ®ex);
103 if (s)
104 return 1;
106 /* The following example is derived from a problem report
107 against gawk from Jorge Stolfi <stolfi@ic.unicamp.br>. */
108 memset (®ex, 0, sizeof regex);
109 s = re_compile_pattern ("[an\371]*n", 7, ®ex);
110 if (s)
111 return 1;
113 /* This should match, but does not for glibc-2.2.1. */
114 if (re_match (®ex, "an", 2, 0, ®s) != 2)
115 return 1;
117 memset (®ex, 0, sizeof regex);
118 s = re_compile_pattern ("x", 1, ®ex);
119 if (s)
120 return 1;
122 /* glibc-2.2.93 does not work with a negative RANGE argument. */
123 if (re_search (®ex, "wxy", 3, 2, -2, ®s) != 1)
124 return 1;
126 /* The version of regex.c in older versions of gnulib
127 ignored RE_ICASE. Detect that problem too. */
128 re_set_syntax (RE_SYNTAX_EMACS | RE_ICASE);
129 memset (®ex, 0, sizeof regex);
130 s = re_compile_pattern ("x", 1, ®ex);
131 if (s)
132 return 1;
134 if (re_search (®ex, "WXY", 3, 0, 3, ®s) < 0)
135 return 1;
137 /* Catch a bug reported by Vin Shelton in
138 http://lists.gnu.org/archive/html/bug-coreutils/2007-06/msg00089.html
139 */
140 re_set_syntax (RE_SYNTAX_POSIX_BASIC
141 & ~RE_CONTEXT_INVALID_DUP
142 & ~RE_NO_EMPTY_RANGES);
143 memset (®ex, 0, sizeof regex);
144 s = re_compile_pattern ("[[:alnum:]_-]\\\\+$", 16, ®ex);
145 if (s)
146 return 1;
148 /* REG_STARTEND was added to glibc on 2004-01-15.
149 Reject older versions. */
150 if (! REG_STARTEND)
151 return 1;
153 /* Reject hosts whose regoff_t values are too narrow.
154 These include glibc 2.3.5 on hosts with 64-bit ptrdiff_t
155 and 32-bit int. */
156 if (sizeof (regoff_t) < sizeof (ptrdiff_t)
157 || sizeof (regoff_t) < sizeof (ssize_t))
158 return 1;
160 return 0;]])],
161 [gl_cv_func_re_compile_pattern_working=yes],
162 [gl_cv_func_re_compile_pattern_working=no],
163 dnl When crosscompiling, assume it is not working.
164 [gl_cv_func_re_compile_pattern_working=no])])
165 case $gl_cv_func_re_compile_pattern_working in #(
166 yes) ac_use_included_regex=no;; #(
167 no) ac_use_included_regex=yes;;
168 esac
169 ;;
170 *) AC_MSG_ERROR([Invalid value for --with-included-regex: $with_included_regex])
171 ;;
172 esac
174 if test $ac_use_included_regex = yes; then
175 AC_DEFINE([_REGEX_LARGE_OFFSETS], [1],
176 [Define if you want regoff_t to be at least as wide POSIX requires.])
177 AC_DEFINE([re_syntax_options], [rpl_re_syntax_options],
178 [Define to rpl_re_syntax_options if the replacement should be used.])
179 AC_DEFINE([re_set_syntax], [rpl_re_set_syntax],
180 [Define to rpl_re_set_syntax if the replacement should be used.])
181 AC_DEFINE([re_compile_pattern], [rpl_re_compile_pattern],
182 [Define to rpl_re_compile_pattern if the replacement should be used.])
183 AC_DEFINE([re_compile_fastmap], [rpl_re_compile_fastmap],
184 [Define to rpl_re_compile_fastmap if the replacement should be used.])
185 AC_DEFINE([re_search], [rpl_re_search],
186 [Define to rpl_re_search if the replacement should be used.])
187 AC_DEFINE([re_search_2], [rpl_re_search_2],
188 [Define to rpl_re_search_2 if the replacement should be used.])
189 AC_DEFINE([re_match], [rpl_re_match],
190 [Define to rpl_re_match if the replacement should be used.])
191 AC_DEFINE([re_match_2], [rpl_re_match_2],
192 [Define to rpl_re_match_2 if the replacement should be used.])
193 AC_DEFINE([re_set_registers], [rpl_re_set_registers],
194 [Define to rpl_re_set_registers if the replacement should be used.])
195 AC_DEFINE([re_comp], [rpl_re_comp],
196 [Define to rpl_re_comp if the replacement should be used.])
197 AC_DEFINE([re_exec], [rpl_re_exec],
198 [Define to rpl_re_exec if the replacement should be used.])
199 AC_DEFINE([regcomp], [rpl_regcomp],
200 [Define to rpl_regcomp if the replacement should be used.])
201 AC_DEFINE([regexec], [rpl_regexec],
202 [Define to rpl_regexec if the replacement should be used.])
203 AC_DEFINE([regerror], [rpl_regerror],
204 [Define to rpl_regerror if the replacement should be used.])
205 AC_DEFINE([regfree], [rpl_regfree],
206 [Define to rpl_regfree if the replacement should be used.])
207 AC_LIBOBJ([regex])
208 gl_PREREQ_REGEX
209 fi
210 ])
212 # Prerequisites of lib/regex.c and lib/regex_internal.c.
213 AC_DEFUN([gl_PREREQ_REGEX],
214 [
215 AC_REQUIRE([AC_USE_SYSTEM_EXTENSIONS])
216 AC_REQUIRE([AC_C_RESTRICT])
217 AC_REQUIRE([AC_TYPE_MBSTATE_T])
218 AC_CHECK_HEADERS([libintl.h])
219 AC_CHECK_FUNCS_ONCE([isblank iswctype wcscoll])
220 AC_CHECK_DECLS([isblank], [], [], [#include <ctype.h>])
221 ])