Imported Upstream version 0.63.0

[hcoop/debian/courier-authlib.git] / unicode / mkultcase.pl

diff --git a/unicode/mkultcase.pl b/unicode/mkultcase.pl
new file mode 100644 (file)
index 0000000..cd336a0
--- /dev/null
+++ b/unicode/mkultcase.pl
@@ -0,0 +1,118 @@
+# Copyright 2000-2004 Double Precision, Inc.
+# See COPYING for distribution information.
+#
+# $Id: mkultcase.pl,v 1.3 2004/02/01 06:09:09 mrsam Exp $
+#
+# Generate unicode upper/lower/titlecase translations.
+
+print '/*
+** Copyright 2000-2004 Double Precision, Inc.
+** See COPYING for distribution information.
+**
+** $Id: mkultcase.pl,v 1.3 2004/02/01 06:09:09 mrsam Exp $
+*/
+
+#include "unicode.h"
+
+';
+
+my $set=shift;
+
+open (U, "UnicodeData.txt") || die "$!\n";
+
+while (<U>)
+{
+       chomp;
+
+my @fields= split /;/;
+
+my ($code, $uc, $lc, $tc);
+
+       $code="0x$fields[0]";
+       eval "\$code=$code;";
+
+       $uc=$fields[12];
+       if ($uc ne "")
+       {
+               eval "\$uc=0x$uc;";
+               $UC{$code}=$uc;
+               $FLAG{$code}=1;
+       }
+
+       $lc=$fields[13];
+       if ($lc ne "")
+       {
+               eval "\$lc=0x$lc;";
+               $LC{$code}=$lc;
+               $FLAG{$code}=1;
+       }
+
+       $tc=$fields[14];
+       if ($tc ne "")
+       {
+               eval "\$tc=0x$tc;";
+               $TC{$code}=$tc;
+               $FLAG{$code}=1;
+       }
+}
+
+close(U);
+
+my $tabsize=1024;
+
+grep ($bucket[ $_ % $tabsize ] .= "$_\n", keys %FLAG);
+
+my $maxcnt=0;
+
+for ($i=0; $i < $tabsize; $i++)
+{
+    my $cnt=0;
+
+    grep ( ++$cnt, split (/\n/, $bucket[$i]));
+
+    $maxcnt=$cnt if $cnt > $maxcnt;
+}
+
+print "const unsigned unicode_case_hash=$tabsize;\n";
+print "/* unicode_case_maxbucket="
+    . ($maxcnt+2) . "*/\n";
+
+print "const unicode_char unicode_case_tab[][4]={\n";
+
+my $idx=0;
+
+for ($i=0; $i<$tabsize; $i++)
+{
+    $offset[$i]=$idx;
+
+    grep {
+       my $j=$_;
+       my $u=$UC{$j}+0;
+       my $l=$LC{$j}+0;
+       my $t=$TC{$j}+0;
+
+       if ($u || $l || $t)
+       {
+           $u=$j unless $u;
+           $l=$j unless $l;
+           $t=$u unless $t;
+
+           printf("{0x%04x,0x%04x,0x%04x,0x%04x},",$j,$u,$l,$t);
+           print "\n" if ($idx % 4) == 3;
+           ++$idx;
+       }
+    } split(/\n/, $bucket[$i]);
+}
+print "{0,0,0,0}};
+
+const unsigned unicode_case_offset[$tabsize]={
+";
+
+for ($i=0; $i<$tabsize;$i++)
+{
+    printf("%4d", $offset[$i]);
+    print "," if $i < $tabsize-1;
+
+    print "\n" if ($i % 16) == 15;
+}
+print "};\n";