File Coverage

blib/lib/Unicode/Collate/CJK/Korean.pm

Criterion	Covered	Total	%
statement	15	15	100.0
branch	2	2	100.0
condition			n/a
subroutine	4	4	100.0
pod	0	1	0.0
total	21	22	95.4

package Unicode::Collate::CJK::Korean;

use 5.006;
use strict;
use warnings;

our $VERSION = '1.28';

my %jamo2prim = (
    '1100', 0x4011,   '1101', 0x4012,   '1102', 0x4013,   '1103', 0x4014,
    '1105', 0x4016,   '1106', 0x4017,   '1107', 0x4018,   '1109', 0x401A,
    '110A', 0x401B,   '110B', 0x401C,   '110C', 0x401D,   '110E', 0x401F,
    '110F', 0x4020,   '1110', 0x4021,   '1111', 0x4022,   '1112', 0x4023,
    '1161', 0x408F,   '1162', 0x4090,   '1163', 0x4091,   '1165', 0x4093,
    '1166', 0x4094,   '1167', 0x4095,   '1168', 0x4096,   '1169', 0x4097,
    '116A', 0x4098,   '116B', 0x4099,   '116C', 0x409A,   '116D', 0x409B,
    '116E', 0x409C,   '116F', 0x409D,   '1170', 0x409E,   '1171', 0x409F,
    '1172', 0x40A0,   '1173', 0x40A1,   '1174', 0x40A2,   '1175', 0x40A3,
    '11A8', 0x40ED,   '11AB', 0x40F0,   '11AF', 0x40F4,   '11B7', 0x40FC,
    '11B8', 0x40FD,   '11BC', 0x4101,   '11BD', 0x4102,
); # for DUCET v12.1.0

my(%u2e, $prim, $wt);
while () {
    last if /^__END__/;
    my @c = split;
    if (@c == 1 && $c[0] =~ /^[A-D]/) { # hangul
	$c[0] =~ s/^.*://;
	$prim = [ map $jamo2prim{$_}, split /-/, $c[0] ];
	$wt = 0x20;
    } else { # ideographs
	for my $c (@c) {
	    next if !$c;
	    $wt++;
	    $u2e{hex($c)} = [ $wt, @$prim ];
	}
    }
}

sub weightKorean {
    my $u = shift;
    return undef if !exists $u2e{$u};
    my @a = @{ $u2e{$u} };
    my $s = shift @a;
    my $p = shift @a;
    return([ $p, $s, 0x2, $u ], @a);
}

1;

# DATA format
# hangul syllable:jamo-jamo(-jamo)
# equiv. ideographs

__DATA__