Commit | Line | Data |
---|---|---|
86530b38 AT |
1 | package Encode::JP::JIS7; |
2 | use strict; | |
3 | ||
4 | our $VERSION = do { my @r = (q$Revision: 1.8 $ =~ /\d+/g); sprintf "%d."."%02d" x $#r, @r }; | |
5 | ||
6 | use Encode qw(:fallbacks); | |
7 | ||
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; | |
11 | ||
12 | $Encode::Encoding{$name} = | |
13 | bless { | |
14 | Name => $name, | |
15 | h2z => $h2z, | |
16 | jis0212 => $jis0212, | |
17 | } => __PACKAGE__; | |
18 | } | |
19 | ||
20 | use base qw(Encode::Encoding); | |
21 | ||
22 | # we override this to 1 so PerlIO works | |
23 | sub needs_lines { 1 } | |
24 | ||
25 | use Encode::CJKConstants qw(:all); | |
26 | ||
27 | our $DEBUG = 0; | |
28 | ||
29 | # | |
30 | # decode is identical for all 2022 variants | |
31 | # | |
32 | ||
33 | sub decode($$;$) | |
34 | { | |
35 | my ($obj, $str, $chk) = @_; | |
36 | my $residue = ''; | |
37 | if ($chk){ | |
38 | $str =~ s/([^\x00-\x7f].*)$//so; | |
39 | $1 and $residue = $1; | |
40 | } | |
41 | $residue .= jis_euc(\$str); | |
42 | $_[1] = $residue if $chk; | |
43 | return Encode::decode('euc-jp', $str, FB_PERLQQ); | |
44 | } | |
45 | ||
46 | # | |
47 | # encode is different | |
48 | # | |
49 | ||
50 | sub encode($$;$) | |
51 | { | |
52 | require Encode::JP::H2Z; | |
53 | my ($obj, $utf8, $chk) = @_; | |
54 | # empty the input string in the stack so perlio is ok | |
55 | $_[1] = '' if $chk; | |
56 | my ($h2z, $jis0212) = @$obj{qw(h2z jis0212)}; | |
57 | my $octet = Encode::encode('euc-jp', $utf8, FB_PERLQQ) ; | |
58 | $h2z and &Encode::JP::H2Z::h2z(\$octet); | |
59 | euc_jis(\$octet, $jis0212); | |
60 | return $octet; | |
61 | } | |
62 | ||
63 | ||
64 | # JIS<->EUC | |
65 | ||
66 | sub jis_euc { | |
67 | my $r_str = shift; | |
68 | $$r_str =~ s( | |
69 | ($RE{JIS_0212}|$RE{JIS_0208}|$RE{ISO_ASC}|$RE{JIS_KANA}) | |
70 | ([^\e]*) | |
71 | ) | |
72 | { | |
73 | my ($esc, $chunk) = ($1, $2); | |
74 | if ($esc !~ /$RE{ISO_ASC}/o) { | |
75 | $chunk =~ tr/\x21-\x7e/\xa1-\xfe/; | |
76 | if ($esc =~ /$RE{JIS_KANA}/o) { | |
77 | $chunk =~ s/([\xa1-\xdf])/\x8e$1/og; | |
78 | } | |
79 | elsif ($esc =~ /$RE{JIS_0212}/o) { | |
80 | $chunk =~ s/([\xa1-\xfe][\xa1-\xfe])/\x8f$1/og; | |
81 | } | |
82 | } | |
83 | $chunk; | |
84 | }geox; | |
85 | my ($residue) = ($$r_str =~ s/(\e.*)$//so); | |
86 | return $residue; | |
87 | } | |
88 | ||
89 | sub euc_jis{ | |
90 | no warnings qw(uninitialized); | |
91 | my $r_str = shift; | |
92 | my $jis0212 = shift; | |
93 | $$r_str =~ s{ | |
94 | ((?:$RE{EUC_C})+|(?:$RE{EUC_KANA})+|(?:$RE{EUC_0212})+) | |
95 | }{ | |
96 | my $chunk = $1; | |
97 | my $esc = | |
98 | ( $chunk =~ tr/\x8E//d ) ? $ESC{KANA} : | |
99 | ( $chunk =~ tr/\x8F//d ) ? $ESC{JIS_0212} : | |
100 | $ESC{JIS_0208}; | |
101 | if ($esc eq $ESC{JIS_0212} && !$jis0212){ | |
102 | # fallback to '?' | |
103 | $chunk =~ tr/\xA1-\xFE/\x3F/; | |
104 | }else{ | |
105 | $chunk =~ tr/\xA1-\xFE/\x21-\x7E/; | |
106 | } | |
107 | $esc . $chunk . $ESC{ASC}; | |
108 | }geox; | |
109 | $$r_str =~ | |
110 | s/\Q$ESC{ASC}\E | |
111 | (\Q$ESC{KANA}\E|\Q$ESC{JIS_0212}\E|\Q$ESC{JIS_0208}\E)/$1/gox; | |
112 | $$r_str; | |
113 | } | |
114 | ||
115 | 1; | |
116 | __END__ | |
117 | ||
118 | ||
119 | =head1 NAME | |
120 | ||
121 | Encode::JP::JIS7 -- internally used by Encode::JP | |
122 | ||
123 | =cut |