Encode/ISO2022/CP932.pm

require 5.7.3;
package Encode::ISO2022::CP932;
use strict;
use vars qw(%DEFAULT $VERSION);
$VERSION=do{my @r=(q$Revision: 1.2 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};
use base qw(Encode::Encoding);
__PACKAGE__->Define (qw/x-iso2022jp-cp932 CP50220/);

sub encode ($$;$) {
  my ($obj, $str, $chk) = @_;
  $_[1] = '' if $chk;
  ## TODO: implement this!
  $str = Encode::encode ('iso-2022-jp', $str);  ## temporary
  return $str;
}

sub decode ($$;$) {
  my ($obj, $str, $chk) = @_;
  $_[1] = '' if $chk;
  _jis7_to_sjis (\$str);
  return Encode::decode ('cp932', $str);
}

my %_L2S;
sub _jis7_to_sjis (\$) {
  my $s = shift;
  $$s =~ s(
        (\x1B\x24\x28?[\x40\x42]|\x1B\x28[BHIJ])
        ([^\x00-\x20]*)
  ){
    my ($esc, $str) = ($1, $2);
    if ($esc =~ /\x1B\x28I/o) {
      $str =~ tr/\x21-\x7E/\xA1-\xDF???????????????????????????????/;
    } elsif ($esc !~ /\x1B\x28/o) {
      $str =~ s{ ([\x21-\xFF][\x21-\x7E]) }{
        my $s = $1;
        unless ($_L2S{$s}){
          my ($c1, $c2) = unpack 'CC', $s;
          $c2 += ($c1 & 1) ? ($c2 < 0x60 ? 0x1F : 0x20) : 0x7E;
          $c1 = (($c1 - 1) >> 1) + ($c1 < 0x5F ? 0x71 : 0xB1);
          $_L2S{$s} = pack 'CC', $c1, $c2;
        }
        $_L2S{$s};
      }goex;
    }
    $str;
  }goex;
  $$s =~ s{ \x0E([\xA1-\xDF]+)\x0F }{
    $1;
  }goex;
}

1;
__END__

=head1 NAME

Encode::ISO2022::CP932 --- Encode module for ISO/IEC 2022 like
encoding of Microsoft CP932 (Shift JIS)

=head1 DESCRIPTION

Windows Code Page 932, Microsoft version of shift JIS,
is widely used in Japanese PC community.  It is the combination
of JIS character set with non-standard extended characters (CCS)
and "shift JIS" encoding scheme (CES).

Non-PC communities such as un*x users' or Internet mail/news'
use standard JIS character set for CCS and EUC or 7bit 
ISO/IEC 2022 for CES.

This situation makes such stupid converters that
output the charset consist of non-standard CCS and EUC
or 7bit ISO/IEC 2022 CES.

This module supports two such charsets.  One is
C<x-iso2022jp-cp932>, 

=head1 ENCODINGS

=over 4

=item x-iso2022jp-cp932

Microsoft Windows CodePage 50220: A transformation format of
Microsoft Windows CodePage 932 that looks like C<ISO-2022-JP>
(Alias: C<CP50220> (M$))

Note that this coding system does NOT comform to RFC 1468,
JIS standards nor ISO/IEC 2022.  This coding system
SHOULD be used when and ONLY when converting data received
from Microsoft Windows platforms.

=back

=head1 LICENSE

Copyright 2002 Wakaba <w@suika.fam.cx>

This library is free software; you can redistribute it
and/or modify it under the same terms as Perl itself.

=cut

# $Date: $
### CP932.pm ends here
1	require 5.7.3;
2	package Encode::ISO2022::CP932;
3	use strict;
4	use vars qw(%DEFAULT $VERSION);
5	$VERSION=do{my @r=(q$Revision: 1.2 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};
6	use base qw(Encode::Encoding);
7	__PACKAGE__->Define (qw/x-iso2022jp-cp932 CP50220/);
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	$1;
50	}goex;
51	}
52
53	1;
54	__END__
55
56	=head1 NAME
57
58	Encode::ISO2022::CP932 --- Encode module for ISO/IEC 2022 like
59	encoding of Microsoft CP932 (Shift JIS)
60
61	=head1 DESCRIPTION
62
63	Windows Code Page 932, Microsoft version of shift JIS,
64	is widely used in Japanese PC community. It is the combination
65	of JIS character set with non-standard extended characters (CCS)
66	and "shift JIS" encoding scheme (CES).
67
68	Non-PC communities such as un*x users' or Internet mail/news'
69	use standard JIS character set for CCS and EUC or 7bit
70	ISO/IEC 2022 for CES.
71
72	This situation makes such stupid converters that
73	output the charset consist of non-standard CCS and EUC
74	or 7bit ISO/IEC 2022 CES.
75
76	This module supports two such charsets. One is
77	C<x-iso2022jp-cp932>,
78
79	=head1 ENCODINGS
80
81	=over 4
82
83	=item x-iso2022jp-cp932
84
85	Microsoft Windows CodePage 50220: A transformation format of
86	Microsoft Windows CodePage 932 that looks like C<ISO-2022-JP>
87	(Alias: C<CP50220> (M$))
88
89	Note that this coding system does NOT comform to RFC 1468,
90	JIS standards nor ISO/IEC 2022. This coding system
91	SHOULD be used when and ONLY when converting data received
92	from Microsoft Windows platforms.
93
94	=back
95
96	=head1 LICENSE
97
98	Copyright 2002 Wakaba <w@suika.fam.cx>
99
100	This library is free software; you can redistribute it
101	and/or modify it under the same terms as Perl itself.
102
103	=cut
104
105	# $Date: $
106	### CP932.pm ends here