m4/regex.m4

   1 # serial 65
   2
   3 # Copyright (C) 1996-2001, 2003-2014 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
   9 dnl Initially derived from code in GNU grep.
  10 dnl Mostly written by Jim Meyering.
  11
  12 AC_PREREQ([2.50])
  13
  14 AC_DEFUN([gl_REGEX],
  15 [
  16   AC_ARG_WITH([included-regex],
  17     [AS_HELP_STRING([--without-included-regex],
  18                     [don't compile regex; this is the default on systems
  19                      with recent-enough versions of the GNU C Library
  20                      (use with caution on other systems).])])
  21
  22   case $with_included_regex in #(
  23   yes|no) ac_use_included_regex=$with_included_regex
  24         ;;
  25   '')
  26     # If the system regex support is good enough that it passes the
  27     # following run test, then default to *not* using the included regex.c.
  28     # If cross compiling, assume the test would fail and use the included
  29     # regex.c.
  30     AC_CHECK_DECLS_ONCE([alarm])
  31     AC_CHECK_HEADERS_ONCE([malloc.h])
  32     AC_CACHE_CHECK([for working re_compile_pattern],
  33                    [gl_cv_func_re_compile_pattern_working],
  34       [AC_RUN_IFELSE(
  35         [AC_LANG_PROGRAM(
  36           [[#include <regex.h>
  37
  38             #include <locale.h>
  39             #include <limits.h>
  40             #include <string.h>
  41
  42             #if defined M_CHECK_ACTION || HAVE_DECL_ALARM
  43             # include <signal.h>
  44             # include <unistd.h>
  45             #endif
  46
  47             #if HAVE_MALLOC_H
  48             # include <malloc.h>
  49             #endif
  50
  51             #ifdef M_CHECK_ACTION
  52             /* Exit with distinguishable exit code.  */
  53             static void sigabrt_no_core (int sig) { raise (SIGTERM); }
  54             #endif
  55           ]],
  56           [[int result = 0;
  57             static struct re_pattern_buffer regex;
  58             unsigned char folded_chars[UCHAR_MAX + 1];
  59             int i;
  60             const char *s;
  61             struct re_registers regs;
  62
  63             /* Some builds of glibc go into an infinite loop on this
  64                test.  Use alarm to force death, and mallopt to avoid
  65                malloc recursion in diagnosing the corrupted heap. */
  66 #if HAVE_DECL_ALARM
  67             signal (SIGALRM, SIG_DFL);
  68             alarm (2);
  69 #endif
  70 #ifdef M_CHECK_ACTION
  71             signal (SIGABRT, sigabrt_no_core);
  72             mallopt (M_CHECK_ACTION, 2);
  73 #endif
  74
  75             if (setlocale (LC_ALL, "en_US.UTF-8"))
  76               {
  77                 {
  78                   /* http://sourceware.org/ml/libc-hacker/2006-09/msg00008.html
  79                      This test needs valgrind to catch the bug on Debian
  80                      GNU/Linux 3.1 x86, but it might catch the bug better
  81                      on other platforms and it shouldn't hurt to try the
  82                      test here.  */
  83                   static char const pat[] = "insert into";
  84                   static char const data[] =
  85                     "\xFF\0\x12\xA2\xAA\xC4\xB1,K\x12\xC4\xB1*\xACK";
  86                   re_set_syntax (RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE
  87                                  | RE_ICASE);
  88                   memset (&regex, 0, sizeof regex);
  89                   s = re_compile_pattern (pat, sizeof pat - 1, &regex);
  90                   if (s)
  91                     result |= 1;
  92                   else if (re_search (&regex, data, sizeof data - 1,
  93                                       0, sizeof data - 1, &regs)
  94                            != -1)
  95                     result |= 1;
  96                 }
  97
  98                 {
  99                   /* This test is from glibc bug 15078.
 100                      The test case is from Andreas Schwab in
 101                      <http://www.sourceware.org/ml/libc-alpha/2013-01/msg00967.html>.
 102                      */
 103                   static char const pat[] = "[^x]x";
 104                   static char const data[] =
 105                     /* <U1000><U103B><U103D><U1014><U103A><U102F><U1015><U103A> */
 106                     "\xe1\x80\x80"
 107                     "\xe1\x80\xbb"
 108                     "\xe1\x80\xbd"
 109                     "\xe1\x80\x94"
 110                     "\xe1\x80\xba"
 111                     "\xe1\x80\xaf"
 112                     "\xe1\x80\x95"
 113                     "\xe1\x80\xba"
 114                     "x";
 115                   re_set_syntax (0);
 116                   memset (&regex, 0, sizeof regex);
 117                   s = re_compile_pattern (pat, sizeof pat - 1, &regex);
 118                   if (s)
 119                     result |= 1;
 120                   else
 121                     {
 122                       i = re_search (&regex, data, sizeof data - 1,
 123                                      0, sizeof data - 1, 0);
 124                       if (i != 0 && i != 21)
 125                         result |= 1;
 126                     }
 127                 }
 128
 129                 if (! setlocale (LC_ALL, "C"))
 130                   return 1;
 131               }
 132
 133             /* This test is from glibc bug 3957, reported by Andrew Mackey.  */
 134             re_set_syntax (RE_SYNTAX_EGREP | RE_HAT_LISTS_NOT_NEWLINE);
 135             memset (&regex, 0, sizeof regex);
 136             s = re_compile_pattern ("a[^x]b", 6, &regex);
 137             if (s)
 138               result |= 2;
 139             /* This should fail, but succeeds for glibc-2.5.  */
 140             else if (re_search (&regex, "a\nb", 3, 0, 3, &regs) != -1)
 141               result |= 2;
 142
 143             /* This regular expression is from Spencer ere test number 75
 144                in grep-2.3.  */
 145             re_set_syntax (RE_SYNTAX_POSIX_EGREP);
 146             memset (&regex, 0, sizeof regex);
 147             for (i = 0; i <= UCHAR_MAX; i++)
 148               folded_chars[i] = i;
 149             regex.translate = folded_chars;
 150             s = re_compile_pattern ("a[[:@:>@:]]b\n", 11, &regex);
 151             /* This should fail with _Invalid character class name_ error.  */
 152             if (!s)
 153               result |= 4;
 154
 155             /* Ensure that [b-a] is diagnosed as invalid, when
 156                using RE_NO_EMPTY_RANGES. */
 157             re_set_syntax (RE_SYNTAX_POSIX_EGREP | RE_NO_EMPTY_RANGES);
 158             memset (&regex, 0, sizeof regex);
 159             s = re_compile_pattern ("a[b-a]", 6, &regex);
 160             if (s == 0)
 161               result |= 8;
 162
 163             /* This should succeed, but does not for glibc-2.1.3.  */
 164             memset (&regex, 0, sizeof regex);
 165             s = re_compile_pattern ("{1", 2, &regex);
 166             if (s)
 167               result |= 8;
 168
 169             /* The following example is derived from a problem report
 170                against gawk from Jorge Stolfi <stolfi@ic.unicamp.br>.  */
 171             memset (&regex, 0, sizeof regex);
 172             s = re_compile_pattern ("[an\371]*n", 7, &regex);
 173             if (s)
 174               result |= 8;
 175             /* This should match, but does not for glibc-2.2.1.  */
 176             else if (re_match (&regex, "an", 2, 0, &regs) != 2)
 177               result |= 8;
 178
 179             memset (&regex, 0, sizeof regex);
 180             s = re_compile_pattern ("x", 1, &regex);
 181             if (s)
 182               result |= 8;
 183             /* glibc-2.2.93 does not work with a negative RANGE argument.  */
 184             else if (re_search (&regex, "wxy", 3, 2, -2, &regs) != 1)
 185               result |= 8;
 186
 187             /* The version of regex.c in older versions of gnulib
 188                ignored RE_ICASE.  Detect that problem too.  */
 189             re_set_syntax (RE_SYNTAX_EMACS | RE_ICASE);
 190             memset (&regex, 0, sizeof regex);
 191             s = re_compile_pattern ("x", 1, &regex);
 192             if (s)
 193               result |= 16;
 194             else if (re_search (&regex, "WXY", 3, 0, 3, &regs) < 0)
 195               result |= 16;
 196
 197             /* Catch a bug reported by Vin Shelton in
 198                http://lists.gnu.org/archive/html/bug-coreutils/2007-06/msg00089.html
 199                */
 200             re_set_syntax (RE_SYNTAX_POSIX_BASIC
 201                            & ~RE_CONTEXT_INVALID_DUP
 202                            & ~RE_NO_EMPTY_RANGES);
 203             memset (&regex, 0, sizeof regex);
 204             s = re_compile_pattern ("[[:alnum:]_-]\\\\+$", 16, &regex);
 205             if (s)
 206               result |= 32;
 207
 208             /* REG_STARTEND was added to glibc on 2004-01-15.
 209                Reject older versions.  */
 210             if (! REG_STARTEND)
 211               result |= 64;
 212
 213 #if 0
 214             /* It would be nice to reject hosts whose regoff_t values are too
 215                narrow (including glibc on hosts with 64-bit ptrdiff_t and
 216                32-bit int), but we should wait until glibc implements this
 217                feature.  Otherwise, support for equivalence classes and
 218                multibyte collation symbols would always be broken except
 219                when compiling --without-included-regex.   */
 220             if (sizeof (regoff_t) < sizeof (ptrdiff_t)
 221                 || sizeof (regoff_t) < sizeof (ssize_t))
 222               result |= 64;
 223 #endif
 224
 225             return result;
 226           ]])],
 227        [gl_cv_func_re_compile_pattern_working=yes],
 228        [gl_cv_func_re_compile_pattern_working=no],
 229        dnl When crosscompiling, assume it is not working.
 230        [gl_cv_func_re_compile_pattern_working=no])])
 231     case $gl_cv_func_re_compile_pattern_working in #(
 232     yes) ac_use_included_regex=no;; #(
 233     no) ac_use_included_regex=yes;;
 234     esac
 235     ;;
 236   *) AC_MSG_ERROR([Invalid value for --with-included-regex: $with_included_regex])
 237     ;;
 238   esac
 239
 240   if test $ac_use_included_regex = yes; then
 241     AC_DEFINE([_REGEX_INCLUDE_LIMITS_H], [1],
 242       [Define if you want <regex.h> to include <limits.h>, so that it
 243        consistently overrides <limits.h>'s RE_DUP_MAX.])
 244     AC_DEFINE([_REGEX_LARGE_OFFSETS], [1],
 245       [Define if you want regoff_t to be at least as wide POSIX requires.])
 246     AC_DEFINE([re_syntax_options], [rpl_re_syntax_options],
 247       [Define to rpl_re_syntax_options if the replacement should be used.])
 248     AC_DEFINE([re_set_syntax], [rpl_re_set_syntax],
 249       [Define to rpl_re_set_syntax if the replacement should be used.])
 250     AC_DEFINE([re_compile_pattern], [rpl_re_compile_pattern],
 251       [Define to rpl_re_compile_pattern if the replacement should be used.])
 252     AC_DEFINE([re_compile_fastmap], [rpl_re_compile_fastmap],
 253       [Define to rpl_re_compile_fastmap if the replacement should be used.])
 254     AC_DEFINE([re_search], [rpl_re_search],
 255       [Define to rpl_re_search if the replacement should be used.])
 256     AC_DEFINE([re_search_2], [rpl_re_search_2],
 257       [Define to rpl_re_search_2 if the replacement should be used.])
 258     AC_DEFINE([re_match], [rpl_re_match],
 259       [Define to rpl_re_match if the replacement should be used.])
 260     AC_DEFINE([re_match_2], [rpl_re_match_2],
 261       [Define to rpl_re_match_2 if the replacement should be used.])
 262     AC_DEFINE([re_set_registers], [rpl_re_set_registers],
 263       [Define to rpl_re_set_registers if the replacement should be used.])
 264     AC_DEFINE([re_comp], [rpl_re_comp],
 265       [Define to rpl_re_comp if the replacement should be used.])
 266     AC_DEFINE([re_exec], [rpl_re_exec],
 267       [Define to rpl_re_exec if the replacement should be used.])
 268     AC_DEFINE([regcomp], [rpl_regcomp],
 269       [Define to rpl_regcomp if the replacement should be used.])
 270     AC_DEFINE([regexec], [rpl_regexec],
 271       [Define to rpl_regexec if the replacement should be used.])
 272     AC_DEFINE([regerror], [rpl_regerror],
 273       [Define to rpl_regerror if the replacement should be used.])
 274     AC_DEFINE([regfree], [rpl_regfree],
 275       [Define to rpl_regfree if the replacement should be used.])
 276   fi
 277 ])
 278
 279 # Prerequisites of lib/regex.c and lib/regex_internal.c.
 280 AC_DEFUN([gl_PREREQ_REGEX],
 281 [
 282   AC_REQUIRE([AC_USE_SYSTEM_EXTENSIONS])
 283   AC_REQUIRE([AC_C_INLINE])
 284   AC_REQUIRE([AC_C_RESTRICT])
 285   AC_REQUIRE([AC_TYPE_MBSTATE_T])
 286   AC_REQUIRE([gl_EEMALLOC])
 287   AC_REQUIRE([gl_GLIBC21])
 288   AC_CHECK_HEADERS([libintl.h])
 289   AC_CHECK_FUNCS_ONCE([isblank iswctype])
 290   AC_CHECK_DECLS([isblank], [], [], [[#include <ctype.h>]])
 291 ])