|
Packit |
d0f5c2 |
package Encode::JP::JIS7;
|
|
Packit |
d0f5c2 |
use strict;
|
|
Packit |
d0f5c2 |
use warnings;
|
|
Packit |
d0f5c2 |
our $VERSION = do { my @r = ( q$Revision: 2.8 $ =~ /\d+/g ); sprintf "%d." . "%02d" x $#r, @r };
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
use Encode qw(:fallbacks);
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
for my $name ( '7bit-jis', 'iso-2022-jp', 'iso-2022-jp-1' ) {
|
|
Packit |
d0f5c2 |
my $h2z = ( $name eq '7bit-jis' ) ? 0 : 1;
|
|
Packit |
d0f5c2 |
my $jis0212 = ( $name eq 'iso-2022-jp' ) ? 0 : 1;
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
my $obj = bless {
|
|
Packit |
d0f5c2 |
Name => $name,
|
|
Packit |
d0f5c2 |
h2z => $h2z,
|
|
Packit |
d0f5c2 |
jis0212 => $jis0212,
|
|
Packit |
d0f5c2 |
} => __PACKAGE__;
|
|
Packit |
d0f5c2 |
Encode::define_encoding($obj, $name);
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
use parent qw(Encode::Encoding);
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
# we override this to 1 so PerlIO works
|
|
Packit |
d0f5c2 |
sub needs_lines { 1 }
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
use Encode::CJKConstants qw(:all);
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
#
|
|
Packit |
d0f5c2 |
# decode is identical for all 2022 variants
|
|
Packit |
d0f5c2 |
#
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
sub decode($$;$) {
|
|
Packit |
d0f5c2 |
my ( $obj, $str, $chk ) = @_;
|
|
Packit |
d0f5c2 |
return undef unless defined $str;
|
|
Packit |
d0f5c2 |
my $residue = '';
|
|
Packit |
d0f5c2 |
if ($chk) {
|
|
Packit |
d0f5c2 |
$str =~ s/([^\x00-\x7f].*)$//so and $residue = $1;
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
$residue .= jis_euc( \$str );
|
|
Packit |
d0f5c2 |
$_[1] = $residue if $chk;
|
|
Packit |
d0f5c2 |
return Encode::decode( 'euc-jp', $str, FB_PERLQQ );
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
#
|
|
Packit |
d0f5c2 |
# encode is different
|
|
Packit |
d0f5c2 |
#
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
sub encode($$;$) {
|
|
Packit |
d0f5c2 |
require Encode::JP::H2Z;
|
|
Packit |
d0f5c2 |
my ( $obj, $utf8, $chk ) = @_;
|
|
Packit |
d0f5c2 |
return undef unless defined $utf8;
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
# empty the input string in the stack so perlio is ok
|
|
Packit |
d0f5c2 |
$_[1] = '' if $chk;
|
|
Packit |
d0f5c2 |
my ( $h2z, $jis0212 ) = @$obj{qw(h2z jis0212)};
|
|
Packit |
d0f5c2 |
my $octet = Encode::encode( 'euc-jp', $utf8, $chk || 0 );
|
|
Packit |
d0f5c2 |
$h2z and &Encode::JP::H2Z::h2z( \$octet );
|
|
Packit |
d0f5c2 |
euc_jis( \$octet, $jis0212 );
|
|
Packit |
d0f5c2 |
return $octet;
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
#
|
|
Packit |
d0f5c2 |
# cat_decode
|
|
Packit |
d0f5c2 |
#
|
|
Packit |
d0f5c2 |
my $re_scan_jis_g = qr{
|
|
Packit |
d0f5c2 |
\G ( ($RE{JIS_0212}) | $RE{JIS_0208} |
|
|
Packit |
d0f5c2 |
($RE{ISO_ASC}) | ($RE{JIS_KANA}) | )
|
|
Packit |
d0f5c2 |
([^\e]*)
|
|
Packit |
d0f5c2 |
}x;
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
sub cat_decode { # ($obj, $dst, $src, $pos, $trm, $chk)
|
|
Packit |
d0f5c2 |
my ( $obj, undef, undef, $pos, $trm ) = @_; # currently ignores $chk
|
|
Packit |
d0f5c2 |
my ( $rdst, $rsrc, $rpos ) = \@_[ 1, 2, 3 ];
|
|
Packit |
d0f5c2 |
local ${^ENCODING};
|
|
Packit |
d0f5c2 |
use bytes;
|
|
Packit |
d0f5c2 |
my $opos = pos($$rsrc);
|
|
Packit |
d0f5c2 |
pos($$rsrc) = $pos;
|
|
Packit |
d0f5c2 |
while ( $$rsrc =~ /$re_scan_jis_g/gc ) {
|
|
Packit |
d0f5c2 |
my ( $esc, $esc_0212, $esc_asc, $esc_kana, $chunk ) =
|
|
Packit |
d0f5c2 |
( $1, $2, $3, $4, $5 );
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
unless ($chunk) { $esc or last; next; }
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
if ( $esc && !$esc_asc ) {
|
|
Packit |
d0f5c2 |
$chunk =~ tr/\x21-\x7e/\xa1-\xfe/;
|
|
Packit |
d0f5c2 |
if ($esc_kana) {
|
|
Packit |
d0f5c2 |
$chunk =~ s/([\xa1-\xdf])/\x8e$1/og;
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
elsif ($esc_0212) {
|
|
Packit |
d0f5c2 |
$chunk =~ s/([\xa1-\xfe][\xa1-\xfe])/\x8f$1/og;
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
$chunk = Encode::decode( 'euc-jp', $chunk, 0 );
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
elsif ( ( my $npos = index( $chunk, $trm ) ) >= 0 ) {
|
|
Packit |
d0f5c2 |
$$rdst .= substr( $chunk, 0, $npos + length($trm) );
|
|
Packit |
d0f5c2 |
$$rpos += length($esc) + $npos + length($trm);
|
|
Packit |
d0f5c2 |
pos($$rsrc) = $opos;
|
|
Packit |
d0f5c2 |
return 1;
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
$$rdst .= $chunk;
|
|
Packit |
d0f5c2 |
$$rpos = pos($$rsrc);
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
$$rpos = pos($$rsrc);
|
|
Packit |
d0f5c2 |
pos($$rsrc) = $opos;
|
|
Packit |
d0f5c2 |
return '';
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
# JIS<->EUC
|
|
Packit |
d0f5c2 |
my $re_scan_jis = qr{
|
|
Packit |
d0f5c2 |
(?:($RE{JIS_0212})|$RE{JIS_0208}|($RE{ISO_ASC})|($RE{JIS_KANA}))([^\e]*)
|
|
Packit |
d0f5c2 |
}x;
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
sub jis_euc {
|
|
Packit |
d0f5c2 |
local ${^ENCODING};
|
|
Packit |
d0f5c2 |
my $r_str = shift;
|
|
Packit |
d0f5c2 |
$$r_str =~ s($re_scan_jis)
|
|
Packit |
d0f5c2 |
{
|
|
Packit |
d0f5c2 |
my ($esc_0212, $esc_asc, $esc_kana, $chunk) =
|
|
Packit |
d0f5c2 |
($1, $2, $3, $4);
|
|
Packit |
d0f5c2 |
if (!$esc_asc) {
|
|
Packit |
d0f5c2 |
$chunk =~ tr/\x21-\x7e/\xa1-\xfe/;
|
|
Packit |
d0f5c2 |
if ($esc_kana) {
|
|
Packit |
d0f5c2 |
$chunk =~ s/([\xa1-\xdf])/\x8e$1/og;
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
elsif ($esc_0212) {
|
|
Packit |
d0f5c2 |
$chunk =~ s/([\xa1-\xfe][\xa1-\xfe])/\x8f$1/og;
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
$chunk;
|
|
Packit |
d0f5c2 |
}geox;
|
|
Packit |
d0f5c2 |
my ($residue) = ( $$r_str =~ s/(\e.*)$//so );
|
|
Packit |
d0f5c2 |
return $residue;
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
sub euc_jis {
|
|
Packit |
d0f5c2 |
no warnings qw(uninitialized);
|
|
Packit |
d0f5c2 |
local ${^ENCODING};
|
|
Packit |
d0f5c2 |
my $r_str = shift;
|
|
Packit |
d0f5c2 |
my $jis0212 = shift;
|
|
Packit |
d0f5c2 |
$$r_str =~ s{
|
|
Packit |
d0f5c2 |
((?:$RE{EUC_C})+|(?:$RE{EUC_KANA})+|(?:$RE{EUC_0212})+)
|
|
Packit |
d0f5c2 |
}{
|
|
Packit |
d0f5c2 |
my $chunk = $1;
|
|
Packit |
d0f5c2 |
my $esc =
|
|
Packit |
d0f5c2 |
( $chunk =~ tr/\x8E//d ) ? $ESC{KANA} :
|
|
Packit |
d0f5c2 |
( $chunk =~ tr/\x8F//d ) ? $ESC{JIS_0212} :
|
|
Packit |
d0f5c2 |
$ESC{JIS_0208};
|
|
Packit |
d0f5c2 |
if ($esc eq $ESC{JIS_0212} && !$jis0212){
|
|
Packit |
d0f5c2 |
# fallback to '?'
|
|
Packit |
d0f5c2 |
$chunk =~ tr/\xA1-\xFE/\x3F/;
|
|
Packit |
d0f5c2 |
}else{
|
|
Packit |
d0f5c2 |
$chunk =~ tr/\xA1-\xFE/\x21-\x7E/;
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
$esc . $chunk . $ESC{ASC};
|
|
Packit |
d0f5c2 |
}geox;
|
|
Packit |
d0f5c2 |
$$r_str =~ s/\Q$ESC{ASC}\E
|
|
Packit |
d0f5c2 |
(\Q$ESC{KANA}\E|\Q$ESC{JIS_0212}\E|\Q$ESC{JIS_0208}\E)/$1/gox;
|
|
Packit |
d0f5c2 |
$$r_str;
|
|
Packit |
d0f5c2 |
}
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
1;
|
|
Packit |
d0f5c2 |
__END__
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
=head1 NAME
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
Encode::JP::JIS7 -- internally used by Encode::JP
|
|
Packit |
d0f5c2 |
|
|
Packit |
d0f5c2 |
=cut
|