Jcode::CP932っぽいもの

Jcode::CP932っぽいもの

　試しに作ってみた。基本的にはUnicodeとの変換をEncodeのCP932を使うように変更しただけです*1。

　他にJcode-2.0以降に含まれるJcode::_Classicと::Constantsと::Trと::H2Zが必要。

　EncodeのFallbackとかどうにかしたほうがいいんだろうなぁ、とか思いつつ。Jcodeのt/*が通るように調整中。

package Jcode::CP932;
use 5.008001;
our $VERSION = '0.01';
use Carp;

use warnings;
use strict;

use Exporter;
use vars qw(@ISA @EXPORT @EXPORT_OK %EXPORT_TAGS $DEBUG);

require Jcode::_Classic;

@ISA         = qw(Jcode::_Classic Exporter);
@EXPORT      = qw(jcode getcode);
@EXPORT_OK   = qw($RCSID $VERSION $DEBUG);
%EXPORT_TAGS = ( all       => [ @EXPORT, @EXPORT_OK ] );
$DEBUG = 0;

use overload 
    q("") => sub { $_[0]->euc },
    q(==) => sub { overload::StrVal($_[0]) eq overload::StrVal($_[1]) },
    q(.=) => sub { $_[0]->append( $_[1] ) },
    fallback => 1,
    ;

#for my $sub (qw/jcode getcode convert load_module/){
for my $sub (qw/getcode convert/){
    no strict 'refs';
    *{$sub} = \&{'Jcode::_Classic::' . $sub };
}
 *Jcode::_Classic::load_module = sub {};

#for my $enc (qw/sjis jis ucs2 utf8/){
for my $enc (qw/sjis jis/) {
    no strict 'refs';
    *{"euc_" . $enc} = \&{"Jcode::_Classic::" . "euc_" . $enc};
    *{$enc . "_euc"} = \&{"Jcode::_Classic::" . $enc . "_euc"};
}
for my $enc (qw/ucs2 utf8/) {
    no strict 'refs';
    *{"Jcode::_Classic::" . "euc_" . $enc} = \&{"euc_" . $enc};
    *{"Jcode::_Classic::" . $enc . "_euc"} = \&{$enc . "_euc"};
}

sub jcode {
    return Jcode::CP932->new(@_)
}

#### Modified
use Encode;
sub euc_ucs2 {
    my $thingy = shift;
    my $sjis = euc_sjis( $thingy );
    Encode::from_to( $sjis, 'cp932', 'UTF-16BE' );
    $sjis;
}
sub ucs2_euc {
    my $thingy = shift;
    my $r_str = ref $thingy ? $thingy : \$thingy;
    Encode::from_to( $$r_str, 'UTF-16BE', 'cp932' );
    sjis_euc( $r_str );
}
sub euc_utf8 {
    my $thingy = shift;
    my $sjis = euc_sjis( $thingy );
    Encode::from_to( $sjis, 'cp932', 'utf8');
    $sjis;
}
sub utf8_euc {
    my $thingy = shift;
    my $r_str = ref $thingy ? $thingy : \$thingy;
    Encode::is_utf8($$r_str) and utf8::encode($$r_str);
    Encode::from_to( $$r_str, 'utf8', 'cp932' );
    sjis_euc( $r_str );
}

1;

*1：Encode::EUCJPMSがあるかどうか判定して、あるならそっちを使うべきな気もするけど。