#!/usr/bin/perl
#
# Split Unicode Data File
-# (c) 1997--2001 Martin Mares <mj@atrey.karlin.mff.cuni.cz>
+# (c) 1997--2003 Martin Mares <mj@atrey.karlin.mff.cuni.cz>
#
open(I, "unidata/UnicodeData.txt") || die "Unable to open UniCode data file";
open(C, ">misc/u-cat") || die "cat file open";
open(U, ">misc/u-upper") || die "upper file open";
open(L, ">misc/u-lower") || die "lower file open";
+open(G, ">misc/u-ligatures") || die "lig file open";
while (<I>) {
chomp;
(/^$/ || /^#/) && next;
($code,$name,$cat,$comb,$bidir,$decomp,$d0,$d1,$n0,$mirr,$cmt1,$cmt2,$upper,$lower,$title) = split /;/;
$code =~ /^....$/ || next;
- if ($cat =~ /^C/) { $ccat = "_C_CTRL"; }
- elsif ($cat =~ /^Z/) { $ccat = "_C_BLANK"; }
- elsif ($cat =~ /^Ll/) { $ccat = "_C_LOWER"; }
- elsif ($cat =~ /^Lu/) { $ccat = "_C_UPPER"; }
- elsif ($code ge "0030" && $code le "0039") { $ccat = "_C_DIGIT|_C_XDIGIT"; }
- elsif ($code eq "005F") { $ccat = "_C_INNER"; }
+ if ($cat =~ /^C/) { $ccat = "_U_CTRL"; }
+ elsif ($cat =~ /^Z/) { $ccat = "_U_SPACE"; }
+ elsif ($decomp =~ /<compat>/ && $name =~ / LIGATURE /) {
+ $ccat = "_U_LIGATURE";
+ print G "$code\n";
+ } elsif ($cat =~ /^Ll/) { $ccat = "_U_LLOWER"; }
+ elsif ($cat =~ /^Lu/) { $ccat = "_U_LUPPER"; }
+ elsif ($cat =~ /^L/) { $ccat = "_U_LETTER"; }
+ elsif ($code ge "0030" && $code le "0039") { $ccat = "_U_DIGIT | _U_XDIGIT"; }
else { $ccat = ""; }
- if ($code ge "0041" && $code le "0046" || $code ge "0061" && $code le "0066") { $ccat = $ccat . "|_C_XDIGIT"; }
+ if ($code ge "0041" && $code le "0046" || $code ge "0061" && $code le "0066") { $ccat = $ccat . "|_U_XDIGIT"; }
if ($ccat ne "") { print C "$code\t$ccat\n"; }
if ($upper ne "") { print U "$code\t0x$upper\n"; }
if ($lower ne "") { print L "$code\t0x$lower\n"; }
close C;
close U;
close L;
+close G;