Commit | Line | Data |
aae85ceb |
1 | package Encode::JP::JIS7; |
2 | use strict; |
3 | |
6d1c0808 |
4 | our $VERSION = do { my @r = (q$Revision: 1.3 $ =~ /\d+/g); sprintf "%d."."%02d" x $#r, @r }; |
5 | |
6 | use Encode qw(:fallbacks); |
aae85ceb |
7 | |
aae85ceb |
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 | |
12 | $Encode::Encoding{$name} = |
aae85ceb |
13 | bless { |
14 | Name => $name, |
15 | h2z => $h2z, |
16 | jis0212 => $jis0212, |
17 | } => __PACKAGE__; |
18 | } |
19 | |
20 | sub name { shift->{'Name'} } |
6d1c0808 |
21 | |
22 | sub new_sequence { $_[0] } |
23 | |
24 | sub needs_lines { 1 } |
aae85ceb |
25 | |
26 | use Encode::CJKConstants qw(:all); |
27 | |
85982a32 |
28 | our $DEBUG = 0; |
29 | |
aae85ceb |
30 | # |
31 | # decode is identical for all 2022 variants |
32 | # |
33 | |
6d1c0808 |
34 | sub decode($$;$) |
aae85ceb |
35 | { |
36 | my ($obj,$str,$chk) = @_; |
85982a32 |
37 | my $residue = jis_euc(\$str); |
38 | # This is for PerlIO |
39 | $_[1] = $residue if $chk; |
40 | # use perlqq fallback for euc-jp -> utf8 |
6d1c0808 |
41 | return Encode::decode('euc-jp', $str, FB_PERLQQ); |
aae85ceb |
42 | } |
43 | |
44 | # |
45 | # encode is different |
46 | # |
47 | |
6d1c0808 |
48 | sub encode($$;$) |
aae85ceb |
49 | { |
50 | require Encode::JP::H2Z; |
85982a32 |
51 | my ($obj, $utf8, $chk) = @_; |
52 | # empty the input string in the stack so perlio is ok |
53 | $_[1] = '' if $chk; |
aae85ceb |
54 | my ($h2z, $jis0212) = @$obj{qw(h2z jis0212)}; |
6d1c0808 |
55 | my $octet = Encode::encode('euc-jp', $utf8, FB_PERLQQ) ; |
85982a32 |
56 | $h2z and &Encode::JP::H2Z::h2z(\$octet); |
57 | euc_jis(\$octet, $jis0212); |
58 | return $octet; |
aae85ceb |
59 | } |
60 | |
61 | |
62 | # JIS<->EUC |
63 | |
64 | sub jis_euc { |
65 | my $r_str = shift; |
66 | $$r_str =~ s( |
67 | ($RE{JIS_0212}|$RE{JIS_0208}|$RE{ISO_ASC}|$RE{JIS_KANA}) |
68 | ([^\e]*) |
69 | ) |
70 | { |
85982a32 |
71 | my ($esc, $chunk) = ($1, $2); |
aae85ceb |
72 | if ($esc !~ /$RE{ISO_ASC}/o) { |
85982a32 |
73 | $chunk =~ tr/\x21-\x7e/\xa1-\xfe/; |
aae85ceb |
74 | if ($esc =~ /$RE{JIS_KANA}/o) { |
85982a32 |
75 | $chunk =~ s/([\xa1-\xdf])/\x8e$1/og; |
aae85ceb |
76 | } |
77 | elsif ($esc =~ /$RE{JIS_0212}/o) { |
85982a32 |
78 | $chunk =~ s/([\xa1-\xfe][\xa1-\xfe])/\x8f$1/og; |
aae85ceb |
79 | } |
80 | } |
85982a32 |
81 | $chunk; |
aae85ceb |
82 | }geox; |
85982a32 |
83 | my ($residue) = ($$r_str =~ s/(\e.*)$//so); |
84 | return $residue; |
aae85ceb |
85 | } |
86 | |
87 | sub euc_jis{ |
88 | my $r_str = shift; |
89 | my $jis0212 = shift; |
90 | $$r_str =~ s{ |
91 | ((?:$RE{EUC_C})+|(?:$RE{EUC_KANA})+|(?:$RE{EUC_0212})+) |
92 | }{ |
85982a32 |
93 | my $chunk = $1; |
6d1c0808 |
94 | my $esc = |
85982a32 |
95 | ( $chunk =~ tr/\x8E//d ) ? $ESC{KANA} : |
96 | ( $chunk =~ tr/\x8F//d ) ? $ESC{JIS_0212} : |
aae85ceb |
97 | $ESC{JIS_0208}; |
98 | if ($esc eq $ESC{JIS_0212} && !$jis0212){ |
99 | # fallback to '?' |
85982a32 |
100 | $chunk =~ tr/\xA1-\xFE/\x3F/; |
aae85ceb |
101 | }else{ |
85982a32 |
102 | $chunk =~ tr/\xA1-\xFE/\x21-\x7E/; |
aae85ceb |
103 | } |
85982a32 |
104 | $esc . $chunk . $ESC{ASC}; |
aae85ceb |
105 | }geox; |
106 | $$r_str =~ |
107 | s/\Q$ESC{ASC}\E |
108 | (\Q$ESC{KANA}\E|\Q$ESC{JIS_0212}\E|\Q$ESC{JIS_0208}\E)/$1/gox; |
109 | $$r_str; |
110 | } |
111 | |
112 | 1; |
113 | __END__ |
114 | |
115 | |
116 | =head1 NAME |
117 | |
118 | Encode::JP::JIS7 -- internally used by Encode::JP |
119 | |
120 | =cut |