| 1 |
require 5.8.0; |
| 2 |
package Encode::ISO2022::CP932; |
| 3 |
use strict; |
| 4 |
use vars qw(%DEFAULT $VERSION); |
| 5 |
$VERSION=do{my @r=(q$Revision: 1.1 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r}; |
| 6 |
use base qw(Encode::Encoding); |
| 7 |
__PACKAGE__->Define (qw/x-iso2022jp-cp932/); |
| 8 |
|
| 9 |
sub encode ($$;$) { |
| 10 |
my ($obj, $str, $chk) = @_; |
| 11 |
$_[1] = '' if $chk; |
| 12 |
## TODO: implement this! |
| 13 |
$str = Encode::encode ('iso-2022-jp', $str); ## temporary |
| 14 |
return $str; |
| 15 |
} |
| 16 |
|
| 17 |
sub decode ($$;$) { |
| 18 |
my ($obj, $str, $chk) = @_; |
| 19 |
$_[1] = '' if $chk; |
| 20 |
_jis7_to_sjis (\$str); |
| 21 |
return Encode::decode ('cp932', $str); |
| 22 |
} |
| 23 |
|
| 24 |
my %_L2S; |
| 25 |
sub _jis7_to_sjis (\$) { |
| 26 |
my $s = shift; |
| 27 |
$$s =~ s( |
| 28 |
(\x1B\x24\x28?[\x40\x42]|\x1B\x28[BHIJ]) |
| 29 |
([^\x00-\x20]*) |
| 30 |
){ |
| 31 |
my ($esc, $str) = ($1, $2); |
| 32 |
if ($esc =~ /\x1B\x28I/o) { |
| 33 |
$str =~ tr/\x21-\x7E/\xA1-\xDF???????????????????????????????/; |
| 34 |
} elsif ($esc !~ /\x1B\x28/o) { |
| 35 |
$str =~ s{ ([\x21-\xFF][\x21-\x7E]) }{ |
| 36 |
my $s = $1; |
| 37 |
unless ($_L2S{$s}){ |
| 38 |
my ($c1, $c2) = unpack 'CC', $s; |
| 39 |
$c2 += ($c1 & 1) ? ($c2 < 0x60 ? 0x1F : 0x20) : 0x7E; |
| 40 |
$c1 = (($c1 - 1) >> 1) + ($c1 < 0x5F ? 0x71 : 0xB1); |
| 41 |
$_L2S{$s} = pack 'CC', $c1, $c2; |
| 42 |
} |
| 43 |
$_L2S{$s}; |
| 44 |
}goex; |
| 45 |
} |
| 46 |
$str; |
| 47 |
}goex; |
| 48 |
$$s =~ s{ \x0E([\xA1-\xDF]+)\x0F }{ |
| 49 |
$1; |
| 50 |
}goex; |
| 51 |
} |
| 52 |
|
| 53 |
1; |
| 54 |
__END__ |
| 55 |
|
| 56 |
=head1 NAME |
| 57 |
|
| 58 |
Encode::ISO2022::CP932 --- Encode module for ISO/IEC 2022 like |
| 59 |
encoding of Microsoft CP932 (Shift JIS) |
| 60 |
|
| 61 |
=head1 DESCRIPTION |
| 62 |
|
| 63 |
Windows Code Page 932, Microsoft version of shift JIS, |
| 64 |
is widely used in Japanese PC community. It is the combination |
| 65 |
of JIS character set with non-standard extended characters (CCS) |
| 66 |
and "shift JIS" encoding scheme (CES). |
| 67 |
|
| 68 |
Non-PC communities such as un*x users' or Internet mail/news' |
| 69 |
use standard JIS character set for CCS and EUC or 7bit |
| 70 |
ISO/IEC 2022 for CES. |
| 71 |
|
| 72 |
This situation makes such stupid converters that |
| 73 |
output the charset consist of non-standard CCS and EUC |
| 74 |
or 7bit ISO/IEC 2022 CES. |
| 75 |
|
| 76 |
This module supports two such charsets. One is |
| 77 |
C<x-iso2022jp-cp932>, |
| 78 |
|
| 79 |
=head1 LICENSE |
| 80 |
|
| 81 |
Copyright 2002 wakaba <w@suika.fam.cx> |
| 82 |
|
| 83 |
This library is free software; you can redistribute it |
| 84 |
and/or modify it under the same terms as Perl itself. |
| 85 |
|
| 86 |
=head1 CHANGE |
| 87 |
|
| 88 |
See F<ChangeLog>. |
| 89 |
$Date: 2002/07/21 05:49:51 $ |
| 90 |
|
| 91 |
=cut |
Parent Directory
| 
Revision Log