Commit | Line | Data |
aae85ceb |
1 | package Encode::JP::JIS7; |
2 | use strict; |
3 | |
151b5d36 |
4 | our $VERSION = do { my @r = (q$Revision: 1.10 $ =~ /\d+/g); sprintf "%d."."%02d" x $#r, @r }; |
6d1c0808 |
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 | |
10c5ecbb |
20 | use base qw(Encode::Encoding); |
6d1c0808 |
21 | |
10c5ecbb |
22 | # we override this to 1 so PerlIO works |
6d1c0808 |
23 | sub needs_lines { 1 } |
aae85ceb |
24 | |
25 | use Encode::CJKConstants qw(:all); |
26 | |
85982a32 |
27 | our $DEBUG = 0; |
28 | |
aae85ceb |
29 | # |
30 | # decode is identical for all 2022 variants |
31 | # |
32 | |
6d1c0808 |
33 | sub decode($$;$) |
aae85ceb |
34 | { |
af1f55d9 |
35 | my ($obj, $str, $chk) = @_; |
36 | my $residue = ''; |
37 | if ($chk){ |
38 | $str =~ s/([^\x00-\x7f].*)$//so; |
39 | $1 and $residue = $1; |
40 | } |
41 | $residue .= jis_euc(\$str); |
85982a32 |
42 | $_[1] = $residue if $chk; |
6d1c0808 |
43 | return Encode::decode('euc-jp', $str, FB_PERLQQ); |
aae85ceb |
44 | } |
45 | |
46 | # |
47 | # encode is different |
48 | # |
49 | |
6d1c0808 |
50 | sub encode($$;$) |
aae85ceb |
51 | { |
52 | require Encode::JP::H2Z; |
85982a32 |
53 | my ($obj, $utf8, $chk) = @_; |
54 | # empty the input string in the stack so perlio is ok |
55 | $_[1] = '' if $chk; |
aae85ceb |
56 | my ($h2z, $jis0212) = @$obj{qw(h2z jis0212)}; |
6d1c0808 |
57 | my $octet = Encode::encode('euc-jp', $utf8, FB_PERLQQ) ; |
85982a32 |
58 | $h2z and &Encode::JP::H2Z::h2z(\$octet); |
59 | euc_jis(\$octet, $jis0212); |
60 | return $octet; |
aae85ceb |
61 | } |
62 | |
220e2d4e |
63 | # |
64 | # cat_decode |
65 | # |
66 | my $re_scan_jis_g = qr{ |
67 | \G ( ($RE{JIS_0212}) | $RE{JIS_0208} | |
68 | ($RE{ISO_ASC}) | ($RE{JIS_KANA}) | ) |
69 | ([^\e]*) |
70 | }x; |
71 | sub cat_decode { # ($obj, $dst, $src, $pos, $trm, $chk) |
72 | my ($obj, undef, undef, $pos, $trm) = @_; # currently ignores $chk |
73 | my ($rdst, $rsrc, $rpos) = \@_[1,2,3]; |
74 | local ${^ENCODING}; |
75 | use bytes; |
76 | my $opos = pos($$rsrc); |
77 | pos($$rsrc) = $pos; |
78 | while ($$rsrc =~ /$re_scan_jis_g/gc) { |
79 | my ($esc, $esc_0212, $esc_asc, $esc_kana, $chunk) = |
80 | ($1, $2, $3, $4, $5); |
81 | |
82 | unless ($chunk) { $esc or last; next; } |
83 | |
84 | if ($esc && !$esc_asc) { |
85 | $chunk =~ tr/\x21-\x7e/\xa1-\xfe/; |
86 | if ($esc_kana) { |
87 | $chunk =~ s/([\xa1-\xdf])/\x8e$1/og; |
88 | } elsif ($esc_0212) { |
89 | $chunk =~ s/([\xa1-\xfe][\xa1-\xfe])/\x8f$1/og; |
90 | } |
91 | $chunk = Encode::decode('euc-jp', $chunk, 0); |
92 | } |
93 | elsif ((my $npos = index($chunk, $trm)) >= 0) { |
94 | $$rdst .= substr($chunk, 0, $npos + length($trm)); |
95 | $$rpos += length($esc) + $npos + length($trm); |
96 | pos($$rsrc) = $opos; |
97 | return 1; |
98 | } |
99 | $$rdst .= $chunk; |
100 | $$rpos = pos($$rsrc); |
101 | } |
102 | $$rpos = pos($$rsrc); |
103 | pos($$rsrc) = $opos; |
104 | return ''; |
105 | } |
aae85ceb |
106 | |
107 | # JIS<->EUC |
220e2d4e |
108 | my $re_scan_jis = qr{ |
b536bf57 |
109 | (?:($RE{JIS_0212})|$RE{JIS_0208}|($RE{ISO_ASC})|($RE{JIS_KANA}))([^\e]*) |
110 | }x; |
aae85ceb |
111 | |
112 | sub jis_euc { |
b536bf57 |
113 | local ${^ENCODING}; |
aae85ceb |
114 | my $r_str = shift; |
b536bf57 |
115 | $$r_str =~ s($re_scan_jis) |
aae85ceb |
116 | { |
b536bf57 |
117 | my ($esc_0212, $esc_asc, $esc_kana, $chunk) = |
118 | ($1, $2, $3, $4); |
119 | if (!$esc_asc) { |
85982a32 |
120 | $chunk =~ tr/\x21-\x7e/\xa1-\xfe/; |
b536bf57 |
121 | if ($esc_kana) { |
85982a32 |
122 | $chunk =~ s/([\xa1-\xdf])/\x8e$1/og; |
aae85ceb |
123 | } |
b536bf57 |
124 | elsif ($esc_0212) { |
85982a32 |
125 | $chunk =~ s/([\xa1-\xfe][\xa1-\xfe])/\x8f$1/og; |
aae85ceb |
126 | } |
127 | } |
85982a32 |
128 | $chunk; |
aae85ceb |
129 | }geox; |
85982a32 |
130 | my ($residue) = ($$r_str =~ s/(\e.*)$//so); |
131 | return $residue; |
aae85ceb |
132 | } |
133 | |
134 | sub euc_jis{ |
0ab8f81e |
135 | no warnings qw(uninitialized); |
aae85ceb |
136 | my $r_str = shift; |
137 | my $jis0212 = shift; |
138 | $$r_str =~ s{ |
139 | ((?:$RE{EUC_C})+|(?:$RE{EUC_KANA})+|(?:$RE{EUC_0212})+) |
140 | }{ |
85982a32 |
141 | my $chunk = $1; |
6d1c0808 |
142 | my $esc = |
85982a32 |
143 | ( $chunk =~ tr/\x8E//d ) ? $ESC{KANA} : |
144 | ( $chunk =~ tr/\x8F//d ) ? $ESC{JIS_0212} : |
aae85ceb |
145 | $ESC{JIS_0208}; |
146 | if ($esc eq $ESC{JIS_0212} && !$jis0212){ |
147 | # fallback to '?' |
85982a32 |
148 | $chunk =~ tr/\xA1-\xFE/\x3F/; |
aae85ceb |
149 | }else{ |
85982a32 |
150 | $chunk =~ tr/\xA1-\xFE/\x21-\x7E/; |
aae85ceb |
151 | } |
85982a32 |
152 | $esc . $chunk . $ESC{ASC}; |
aae85ceb |
153 | }geox; |
154 | $$r_str =~ |
155 | s/\Q$ESC{ASC}\E |
156 | (\Q$ESC{KANA}\E|\Q$ESC{JIS_0212}\E|\Q$ESC{JIS_0208}\E)/$1/gox; |
157 | $$r_str; |
158 | } |
159 | |
160 | 1; |
161 | __END__ |
162 | |
163 | |
164 | =head1 NAME |
165 | |
166 | Encode::JP::JIS7 -- internally used by Encode::JP |
167 | |
168 | =cut |