regexp: Fix generation of _unicode_mapping.c.

* configure.ac (AWK_HEX_NUMBER_OPTION): Detect GNU Awk.
* regexp/Makefile.am: Use AWK_HEX_NUMBER_OPTION.
* regexp/parse-unidata.awk: Don't use strtonum.

GnuPG-bug-id: 4915
Signed-off-by: NIIBE Yutaka <gniibe@fsij.org>

configure.ac

@@ -633,6 +633,16 @@ AC_ISC_POSIX
 AC_SYS_LARGEFILE
 
 
+# GNU AWK requires -n option to interpret "0xHH" as a number
+if $AWK 'BEGIN { if (PROCINFO@<:@"version"@:>@) exit 1 }'; then
+  AWK_HEX_NUMBER_OPTION=''
+  AC_MSG_NOTICE([awk with no option for hexadecimal])
+else
+  AWK_HEX_NUMBER_OPTION='-n'
+  AC_MSG_NOTICE([awk with an option -n for hexadecimal])
+fi
+AC_SUBST(AWK_HEX_NUMBER_OPTION)
+
 # We need to compile and run a program on the build machine.  A
 # comment in libgpg-error says that the AC_PROG_CC_FOR_BUILD macro in
 # the AC archive is broken for autoconf 2.57.  Given that there is no

regexp/Makefile.am

@@ -34,5 +34,6 @@ BUILT_SOURCES = _unicode_mapping.c
 MAINTAINERCLEANFILES = _unicode_mapping.c
 
 _unicode_mapping.c: parse-unidata.awk UnicodeData.txt
-	$(AWK) -f $(srcdir)/parse-unidata.awk $(srcdir)/UnicodeData.txt >$@
+	$(AWK) $(AWK_HEX_NUMBER_OPTION) -f $(srcdir)/parse-unidata.awk \
+	       $(srcdir)/UnicodeData.txt >$@
 endif

regexp/parse-unidata.awk

@@ -32,17 +32,17 @@ BEGIN {
 }
 
 {
-    code = strtonum(("0x" $1))
+    code = int("0x" $1)
     name = $2
     class = $3
     upper = $13
     lower = $14
     title = $15
 
-    if (code <= 0x7f) {
+    if (code <= 127) {
 	next
     }
-    if (code > 0xffff) {
+    if (code > 65535) {
 	next
     }
     if ($3 !~ /^L.*/) {