Commit | Line | Data |
aae85ceb |
1 | package Encode::JP::JIS7; |
2 | use strict; |
656ebd29 |
3 | use warnings; |
4 | our $VERSION = do { my @r = ( q$Revision: 2.2 $ =~ /\d+/g ); sprintf "%d." . "%02d" x $#r, @r }; |
6d1c0808 |
5 | |
6 | use Encode qw(:fallbacks); |
aae85ceb |
7 | |
d1256cb1 |
8 | for my $name ( '7bit-jis', 'iso-2022-jp', 'iso-2022-jp-1' ) { |
9 | my $h2z = ( $name eq '7bit-jis' ) ? 0 : 1; |
10 | my $jis0212 = ( $name eq 'iso-2022-jp' ) ? 0 : 1; |
6d1c0808 |
11 | |
d1256cb1 |
12 | $Encode::Encoding{$name} = bless { |
13 | Name => $name, |
14 | h2z => $h2z, |
15 | jis0212 => $jis0212, |
16 | } => __PACKAGE__; |
aae85ceb |
17 | } |
18 | |
10c5ecbb |
19 | use base qw(Encode::Encoding); |
6d1c0808 |
20 | |
10c5ecbb |
21 | # we override this to 1 so PerlIO works |
6d1c0808 |
22 | sub needs_lines { 1 } |
aae85ceb |
23 | |
24 | use Encode::CJKConstants qw(:all); |
25 | |
26 | # |
27 | # decode is identical for all 2022 variants |
28 | # |
29 | |
d1256cb1 |
30 | sub decode($$;$) { |
31 | my ( $obj, $str, $chk ) = @_; |
af1f55d9 |
32 | my $residue = ''; |
d1256cb1 |
33 | if ($chk) { |
34 | $str =~ s/([^\x00-\x7f].*)$//so and $residue = $1; |
af1f55d9 |
35 | } |
d1256cb1 |
36 | $residue .= jis_euc( \$str ); |
85982a32 |
37 | $_[1] = $residue if $chk; |
d1256cb1 |
38 | return Encode::decode( 'euc-jp', $str, FB_PERLQQ ); |
aae85ceb |
39 | } |
40 | |
41 | # |
42 | # encode is different |
43 | # |
44 | |
d1256cb1 |
45 | sub encode($$;$) { |
aae85ceb |
46 | require Encode::JP::H2Z; |
d1256cb1 |
47 | my ( $obj, $utf8, $chk ) = @_; |
48 | |
85982a32 |
49 | # empty the input string in the stack so perlio is ok |
50 | $_[1] = '' if $chk; |
d1256cb1 |
51 | my ( $h2z, $jis0212 ) = @$obj{qw(h2z jis0212)}; |
52 | my $octet = Encode::encode( 'euc-jp', $utf8, FB_PERLQQ ); |
53 | $h2z and &Encode::JP::H2Z::h2z( \$octet ); |
54 | euc_jis( \$octet, $jis0212 ); |
85982a32 |
55 | return $octet; |
aae85ceb |
56 | } |
57 | |
220e2d4e |
58 | # |
59 | # cat_decode |
60 | # |
61 | my $re_scan_jis_g = qr{ |
62 | \G ( ($RE{JIS_0212}) | $RE{JIS_0208} | |
63 | ($RE{ISO_ASC}) | ($RE{JIS_KANA}) | ) |
64 | ([^\e]*) |
65 | }x; |
d1256cb1 |
66 | |
67 | sub cat_decode { # ($obj, $dst, $src, $pos, $trm, $chk) |
68 | my ( $obj, undef, undef, $pos, $trm ) = @_; # currently ignores $chk |
69 | my ( $rdst, $rsrc, $rpos ) = \@_[ 1, 2, 3 ]; |
220e2d4e |
70 | local ${^ENCODING}; |
71 | use bytes; |
72 | my $opos = pos($$rsrc); |
73 | pos($$rsrc) = $pos; |
d1256cb1 |
74 | while ( $$rsrc =~ /$re_scan_jis_g/gc ) { |
75 | my ( $esc, $esc_0212, $esc_asc, $esc_kana, $chunk ) = |
76 | ( $1, $2, $3, $4, $5 ); |
77 | |
78 | unless ($chunk) { $esc or last; next; } |
79 | |
80 | if ( $esc && !$esc_asc ) { |
81 | $chunk =~ tr/\x21-\x7e/\xa1-\xfe/; |
82 | if ($esc_kana) { |
83 | $chunk =~ s/([\xa1-\xdf])/\x8e$1/og; |
84 | } |
85 | elsif ($esc_0212) { |
86 | $chunk =~ s/([\xa1-\xfe][\xa1-\xfe])/\x8f$1/og; |
87 | } |
88 | $chunk = Encode::decode( 'euc-jp', $chunk, 0 ); |
89 | } |
90 | elsif ( ( my $npos = index( $chunk, $trm ) ) >= 0 ) { |
91 | $$rdst .= substr( $chunk, 0, $npos + length($trm) ); |
92 | $$rpos += length($esc) + $npos + length($trm); |
93 | pos($$rsrc) = $opos; |
94 | return 1; |
95 | } |
96 | $$rdst .= $chunk; |
97 | $$rpos = pos($$rsrc); |
220e2d4e |
98 | } |
99 | $$rpos = pos($$rsrc); |
100 | pos($$rsrc) = $opos; |
101 | return ''; |
102 | } |
aae85ceb |
103 | |
104 | # JIS<->EUC |
220e2d4e |
105 | my $re_scan_jis = qr{ |
b536bf57 |
106 | (?:($RE{JIS_0212})|$RE{JIS_0208}|($RE{ISO_ASC})|($RE{JIS_KANA}))([^\e]*) |
107 | }x; |
aae85ceb |
108 | |
109 | sub jis_euc { |
b536bf57 |
110 | local ${^ENCODING}; |
aae85ceb |
111 | my $r_str = shift; |
b536bf57 |
112 | $$r_str =~ s($re_scan_jis) |
aae85ceb |
113 | { |
d1256cb1 |
114 | my ($esc_0212, $esc_asc, $esc_kana, $chunk) = |
115 | ($1, $2, $3, $4); |
116 | if (!$esc_asc) { |
117 | $chunk =~ tr/\x21-\x7e/\xa1-\xfe/; |
118 | if ($esc_kana) { |
119 | $chunk =~ s/([\xa1-\xdf])/\x8e$1/og; |
120 | } |
121 | elsif ($esc_0212) { |
122 | $chunk =~ s/([\xa1-\xfe][\xa1-\xfe])/\x8f$1/og; |
123 | } |
124 | } |
125 | $chunk; |
aae85ceb |
126 | }geox; |
d1256cb1 |
127 | my ($residue) = ( $$r_str =~ s/(\e.*)$//so ); |
85982a32 |
128 | return $residue; |
aae85ceb |
129 | } |
130 | |
d1256cb1 |
131 | sub euc_jis { |
0ab8f81e |
132 | no warnings qw(uninitialized); |
d1256cb1 |
133 | my $r_str = shift; |
aae85ceb |
134 | my $jis0212 = shift; |
135 | $$r_str =~ s{ |
d1256cb1 |
136 | ((?:$RE{EUC_C})+|(?:$RE{EUC_KANA})+|(?:$RE{EUC_0212})+) |
137 | }{ |
138 | my $chunk = $1; |
139 | my $esc = |
140 | ( $chunk =~ tr/\x8E//d ) ? $ESC{KANA} : |
141 | ( $chunk =~ tr/\x8F//d ) ? $ESC{JIS_0212} : |
142 | $ESC{JIS_0208}; |
143 | if ($esc eq $ESC{JIS_0212} && !$jis0212){ |
144 | # fallback to '?' |
145 | $chunk =~ tr/\xA1-\xFE/\x3F/; |
146 | }else{ |
147 | $chunk =~ tr/\xA1-\xFE/\x21-\x7E/; |
148 | } |
149 | $esc . $chunk . $ESC{ASC}; |
150 | }geox; |
151 | $$r_str =~ s/\Q$ESC{ASC}\E |
152 | (\Q$ESC{KANA}\E|\Q$ESC{JIS_0212}\E|\Q$ESC{JIS_0208}\E)/$1/gox; |
aae85ceb |
153 | $$r_str; |
154 | } |
155 | |
156 | 1; |
157 | __END__ |
158 | |
159 | |
160 | =head1 NAME |
161 | |
162 | Encode::JP::JIS7 -- internally used by Encode::JP |
163 | |
164 | =cut |