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 = '1.30';

my %jamo2prim = (
    '1100', 0x4175,   '1101', 0x4176,   '1102', 0x4177,   '1103', 0x4178,
    '1105', 0x417A,   '1106', 0x417B,   '1107', 0x417C,   '1109', 0x417E,
    '110A', 0x417F,   '110B', 0x4180,   '110C', 0x4181,   '110E', 0x4183,
    '110F', 0x4184,   '1110', 0x4185,   '1111', 0x4186,   '1112', 0x4187,
    '1161', 0x41F3,   '1162', 0x41F4,   '1163', 0x41F5,   '1165', 0x41F7,
    '1166', 0x41F8,   '1167', 0x41F9,   '1168', 0x41FA,   '1169', 0x41FB,
    '116A', 0x41FC,   '116B', 0x41FD,   '116C', 0x41FE,   '116D', 0x41FF,
    '116E', 0x4200,   '116F', 0x4201,   '1170', 0x4202,   '1171', 0x4203,
    '1172', 0x4204,   '1173', 0x4205,   '1174', 0x4206,   '1175', 0x4207,
    '11A8', 0x4251,   '11AB', 0x4254,   '11AF', 0x4258,   '11B7', 0x4260,
    '11B8', 0x4261,   '11BC', 0x4265,   '11BD', 0x4266,
); # for DUCET v13.0.0

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__