summaryrefslogtreecommitdiffstats
path: root/gl/m4/regex.m4
diff options
context:
space:
mode:
Diffstat (limited to 'gl/m4/regex.m4')
-rw-r--r--gl/m4/regex.m4198
1 files changed, 198 insertions, 0 deletions
diff --git a/gl/m4/regex.m4 b/gl/m4/regex.m4
new file mode 100644
index 00000000..25da645e
--- /dev/null
+++ b/gl/m4/regex.m4
@@ -0,0 +1,198 @@
1#serial 42
2
3# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2003, 2004, 2005,
4# 2006, 2007 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.
9
10dnl Initially derived from code in GNU grep.
11dnl Mostly written by Jim Meyering.
12
13AC_PREREQ([2.50])
14
15AC_DEFUN([gl_REGEX],
16[
17 AC_CHECK_HEADERS_ONCE([locale.h])
18
19 AC_ARG_WITH([included-regex],
20 [AC_HELP_STRING([--without-included-regex],
21 [don't compile regex; this is the default on
22 systems with recent-enough versions of the GNU C
23 Library (use with caution on other systems)])])
24
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. The first failing regular expression is from `Spencer ere
33 # test #75' in grep-2.3.
34 AC_CACHE_CHECK([for working re_compile_pattern],
35 [gl_cv_func_re_compile_pattern_working],
36 [AC_RUN_IFELSE(
37 [AC_LANG_PROGRAM(
38 [AC_INCLUDES_DEFAULT
39 #if HAVE_LOCALE_H
40 #include <locale.h>
41 #endif
42 #include <limits.h>
43 #include <regex.h>
44 ],
45 [[static struct re_pattern_buffer regex;
46 unsigned char folded_chars[UCHAR_MAX + 1];
47 int i;
48 const char *s;
49 struct re_registers regs;
50
51 #if HAVE_LOCALE_H
52 /* http://sourceware.org/ml/libc-hacker/2006-09/msg00008.html
53 This test needs valgrind to catch the bug on Debian
54 GNU/Linux 3.1 x86, but it might catch the bug better
55 on other platforms and it shouldn't hurt to try the
56 test here. */
57 if (setlocale (LC_ALL, "en_US.UTF-8"))
58 {
59 static char const pat[] = "insert into";
60 static char const data[] =
61 "\xFF\0\x12\xA2\xAA\xC4\xB1,K\x12\xC4\xB1*\xACK";
62 re_set_syntax (RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE
63 | RE_ICASE);
64 memset (&regex, 0, sizeof regex);
65 s = re_compile_pattern (pat, sizeof pat - 1, &regex);
66 if (s)
67 return 1;
68 if (re_search (&regex, data, sizeof data - 1,
69 0, sizeof data - 1, &regs)
70 != -1)
71 return 1;
72 if (! setlocale (LC_ALL, "C"))
73 return 1;
74 }
75 #endif
76
77 re_set_syntax (RE_SYNTAX_POSIX_EGREP);
78 memset (&regex, 0, sizeof (regex));
79 for (i = 0; i <= UCHAR_MAX; i++)
80 folded_chars[i] = i;
81 regex.translate = folded_chars;
82 s = re_compile_pattern ("a[[:@:>@:]]b\n", 11, &regex);
83 /* This should fail with _Invalid character class name_ error. */
84 if (!s)
85 exit (1);
86
87 /* This should succeed, but does not for e.g. glibc-2.1.3. */
88 memset (&regex, 0, sizeof (regex));
89 s = re_compile_pattern ("{1", 2, &regex);
90
91 if (s)
92 exit (1);
93
94 /* The following example is derived from a problem report
95 against gawk from Jorge Stolfi <stolfi@ic.unicamp.br>. */
96 memset (&regex, 0, sizeof (regex));
97 s = re_compile_pattern ("[an\371]*n", 7, &regex);
98 if (s)
99 exit (1);
100
101 /* This should match, but does not for e.g. glibc-2.2.1. */
102 if (re_match (&regex, "an", 2, 0, &regs) != 2)
103 exit (1);
104
105 memset (&regex, 0, sizeof (regex));
106 s = re_compile_pattern ("x", 1, &regex);
107 if (s)
108 exit (1);
109
110 /* The version of regex.c in e.g. GNU libc-2.2.93 did not
111 work with a negative RANGE argument. */
112 if (re_search (&regex, "wxy", 3, 2, -2, &regs) != 1)
113 exit (1);
114
115 /* The version of regex.c in older versions of gnulib
116 ignored RE_ICASE. Detect that problem too. */
117 memset (&regex, 0, sizeof (regex));
118 re_set_syntax (RE_SYNTAX_EMACS | RE_ICASE);
119 s = re_compile_pattern ("x", 1, &regex);
120 if (s)
121 exit (1);
122
123 if (re_search (&regex, "WXY", 3, 0, 3, &regs) < 0)
124 exit (1);
125
126 /* REG_STARTEND was added to glibc on 2004-01-15.
127 Reject older versions. */
128 if (! REG_STARTEND)
129 exit (1);
130
131 /* Reject hosts whose regoff_t values are too narrow.
132 These include glibc 2.3.5 on hosts with 64-bit ptrdiff_t
133 and 32-bit int. */
134 if (sizeof (regoff_t) < sizeof (ptrdiff_t)
135 || sizeof (regoff_t) < sizeof (ssize_t))
136 exit (1);
137
138 exit (0);]])],
139 [gl_cv_func_re_compile_pattern_working=yes],
140 [gl_cv_func_re_compile_pattern_working=no],
141 dnl When crosscompiling, assume it is not working.
142 [gl_cv_func_re_compile_pattern_working=no])])
143 case $gl_cv_func_re_compile_pattern_working in #(
144 yes) ac_use_included_regex=no;; #(
145 no) ac_use_included_regex=yes;;
146 esac
147 ;;
148 *) AC_MSG_ERROR([Invalid value for --with-included-regex: $with_included_regex])
149 ;;
150 esac
151
152 if test $ac_use_included_regex = yes; then
153 AC_DEFINE([_REGEX_LARGE_OFFSETS], 1,
154 [Define if you want regoff_t to be at least as wide POSIX requires.])
155 AC_DEFINE([re_syntax_options], [rpl_re_syntax_options],
156 [Define to rpl_re_syntax_options if the replacement should be used.])
157 AC_DEFINE([re_set_syntax], [rpl_re_set_syntax],
158 [Define to rpl_re_set_syntax if the replacement should be used.])
159 AC_DEFINE([re_compile_pattern], [rpl_re_compile_pattern],
160 [Define to rpl_re_compile_pattern if the replacement should be used.])
161 AC_DEFINE([re_compile_fastmap], [rpl_re_compile_fastmap],
162 [Define to rpl_re_compile_fastmap if the replacement should be used.])
163 AC_DEFINE([re_search], [rpl_re_search],
164 [Define to rpl_re_search if the replacement should be used.])
165 AC_DEFINE([re_search_2], [rpl_re_search_2],
166 [Define to rpl_re_search_2 if the replacement should be used.])
167 AC_DEFINE([re_match], [rpl_re_match],
168 [Define to rpl_re_match if the replacement should be used.])
169 AC_DEFINE([re_match_2], [rpl_re_match_2],
170 [Define to rpl_re_match_2 if the replacement should be used.])
171 AC_DEFINE([re_set_registers], [rpl_re_set_registers],
172 [Define to rpl_re_set_registers if the replacement should be used.])
173 AC_DEFINE([re_comp], [rpl_re_comp],
174 [Define to rpl_re_comp if the replacement should be used.])
175 AC_DEFINE([re_exec], [rpl_re_exec],
176 [Define to rpl_re_exec if the replacement should be used.])
177 AC_DEFINE([regcomp], [rpl_regcomp],
178 [Define to rpl_regcomp if the replacement should be used.])
179 AC_DEFINE([regexec], [rpl_regexec],
180 [Define to rpl_regexec if the replacement should be used.])
181 AC_DEFINE([regerror], [rpl_regerror],
182 [Define to rpl_regerror if the replacement should be used.])
183 AC_DEFINE([regfree], [rpl_regfree],
184 [Define to rpl_regfree if the replacement should be used.])
185 AC_LIBOBJ([regex])
186 gl_PREREQ_REGEX
187 fi
188])
189
190# Prerequisites of lib/regex.c and lib/regex_internal.c.
191AC_DEFUN([gl_PREREQ_REGEX],
192[
193 AC_REQUIRE([AC_GNU_SOURCE])
194 AC_REQUIRE([AC_C_RESTRICT])
195 AC_REQUIRE([AM_LANGINFO_CODESET])
196 AC_CHECK_FUNCS_ONCE([iswctype mbrtowc mempcpy wcrtomb wcscoll])
197 AC_CHECK_DECLS([isblank], [], [], [#include <ctype.h>])
198])