line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Encode::JP::JIS7; |
2
|
14
|
|
|
14
|
|
110
|
use strict; |
|
14
|
|
|
|
|
40
|
|
|
14
|
|
|
|
|
442
|
|
3
|
14
|
|
|
14
|
|
94
|
use warnings; |
|
14
|
|
|
|
|
39
|
|
|
14
|
|
|
|
|
1245
|
|
4
|
|
|
|
|
|
|
our $VERSION = do { my @r = ( q$Revision: 2.7 $ =~ /\d+/g ); sprintf "%d." . "%02d" x $#r, @r }; |
5
|
|
|
|
|
|
|
|
6
|
14
|
|
|
14
|
|
105
|
use Encode qw(:fallbacks); |
|
14
|
|
|
|
|
48
|
|
|
14
|
|
|
|
|
3715
|
|
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
|
|
|
|
|
|
|
my $obj = bless { |
13
|
|
|
|
|
|
|
Name => $name, |
14
|
|
|
|
|
|
|
h2z => $h2z, |
15
|
|
|
|
|
|
|
jis0212 => $jis0212, |
16
|
|
|
|
|
|
|
} => __PACKAGE__; |
17
|
|
|
|
|
|
|
Encode::define_encoding($obj, $name); |
18
|
|
|
|
|
|
|
} |
19
|
|
|
|
|
|
|
|
20
|
14
|
|
|
14
|
|
138
|
use parent qw(Encode::Encoding); |
|
14
|
|
|
|
|
37
|
|
|
14
|
|
|
|
|
174
|
|
21
|
|
|
|
|
|
|
|
22
|
|
|
|
|
|
|
# we override this to 1 so PerlIO works |
23
|
12
|
|
|
12
|
1
|
206
|
sub needs_lines { 1 } |
24
|
|
|
|
|
|
|
|
25
|
14
|
|
|
14
|
|
6635
|
use Encode::CJKConstants qw(:all); |
|
14
|
|
|
|
|
56
|
|
|
14
|
|
|
|
|
6953
|
|
26
|
|
|
|
|
|
|
|
27
|
|
|
|
|
|
|
# |
28
|
|
|
|
|
|
|
# decode is identical for all 2022 variants |
29
|
|
|
|
|
|
|
# |
30
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
sub decode($$;$) { |
32
|
47
|
|
|
47
|
1
|
2120
|
my ( $obj, $str, $chk ) = @_; |
33
|
47
|
100
|
|
|
|
211
|
return undef unless defined $str; |
34
|
44
|
|
|
|
|
130
|
my $residue = ''; |
35
|
44
|
100
|
|
|
|
146
|
if ($chk) { |
36
|
26
|
100
|
|
|
|
1772
|
$str =~ s/([^\x00-\x7f].*)$//so and $residue = $1; |
37
|
|
|
|
|
|
|
} |
38
|
44
|
|
|
|
|
208
|
$residue .= jis_euc( \$str ); |
39
|
44
|
100
|
|
|
|
263
|
$_[1] = $residue if $chk; |
40
|
44
|
|
|
|
|
241
|
return Encode::decode( 'euc-jp', $str, FB_PERLQQ ); |
41
|
|
|
|
|
|
|
} |
42
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
# |
44
|
|
|
|
|
|
|
# encode is different |
45
|
|
|
|
|
|
|
# |
46
|
|
|
|
|
|
|
|
47
|
|
|
|
|
|
|
sub encode($$;$) { |
48
|
1397
|
|
|
1397
|
1
|
18443
|
require Encode::JP::H2Z; |
49
|
1397
|
|
|
|
|
4432
|
my ( $obj, $utf8, $chk ) = @_; |
50
|
1397
|
100
|
|
|
|
4513
|
return undef unless defined $utf8; |
51
|
|
|
|
|
|
|
|
52
|
|
|
|
|
|
|
# empty the input string in the stack so perlio is ok |
53
|
1394
|
100
|
|
|
|
4167
|
$_[1] = '' if $chk; |
54
|
1394
|
|
|
|
|
4118
|
my ( $h2z, $jis0212 ) = @$obj{qw(h2z jis0212)}; |
55
|
1394
|
|
|
|
|
4857
|
my $octet = Encode::encode( 'euc-jp', $utf8, $chk ); |
56
|
1394
|
100
|
|
|
|
6801
|
$h2z and &Encode::JP::H2Z::h2z( \$octet ); |
57
|
1394
|
|
|
|
|
5056
|
euc_jis( \$octet, $jis0212 ); |
58
|
1394
|
|
|
|
|
25071
|
return $octet; |
59
|
|
|
|
|
|
|
} |
60
|
|
|
|
|
|
|
|
61
|
|
|
|
|
|
|
# |
62
|
|
|
|
|
|
|
# cat_decode |
63
|
|
|
|
|
|
|
# |
64
|
|
|
|
|
|
|
my $re_scan_jis_g = qr{ |
65
|
|
|
|
|
|
|
\G ( ($RE{JIS_0212}) | $RE{JIS_0208} | |
66
|
|
|
|
|
|
|
($RE{ISO_ASC}) | ($RE{JIS_KANA}) | ) |
67
|
|
|
|
|
|
|
([^\e]*) |
68
|
|
|
|
|
|
|
}x; |
69
|
|
|
|
|
|
|
|
70
|
|
|
|
|
|
|
sub cat_decode { # ($obj, $dst, $src, $pos, $trm, $chk) |
71
|
0
|
|
|
0
|
1
|
0
|
my ( $obj, undef, undef, $pos, $trm ) = @_; # currently ignores $chk |
72
|
0
|
|
|
|
|
0
|
my ( $rdst, $rsrc, $rpos ) = \@_[ 1, 2, 3 ]; |
73
|
0
|
|
|
|
|
0
|
local ${^ENCODING}; |
74
|
14
|
|
|
14
|
|
122
|
use bytes; |
|
14
|
|
|
|
|
40
|
|
|
14
|
|
|
|
|
116
|
|
75
|
0
|
|
|
|
|
0
|
my $opos = pos($$rsrc); |
76
|
0
|
|
|
|
|
0
|
pos($$rsrc) = $pos; |
77
|
0
|
|
|
|
|
0
|
while ( $$rsrc =~ /$re_scan_jis_g/gc ) { |
78
|
0
|
|
|
|
|
0
|
my ( $esc, $esc_0212, $esc_asc, $esc_kana, $chunk ) = |
79
|
|
|
|
|
|
|
( $1, $2, $3, $4, $5 ); |
80
|
|
|
|
|
|
|
|
81
|
0
|
0
|
|
|
|
0
|
unless ($chunk) { $esc or last; next; } |
|
0
|
0
|
|
|
|
0
|
|
|
0
|
|
|
|
|
0
|
|
82
|
|
|
|
|
|
|
|
83
|
0
|
0
|
0
|
|
|
0
|
if ( $esc && !$esc_asc ) { |
|
|
0
|
|
|
|
|
|
84
|
0
|
|
|
|
|
0
|
$chunk =~ tr/\x21-\x7e/\xa1-\xfe/; |
85
|
0
|
0
|
|
|
|
0
|
if ($esc_kana) { |
|
|
0
|
|
|
|
|
|
86
|
0
|
|
|
|
|
0
|
$chunk =~ s/([\xa1-\xdf])/\x8e$1/og; |
87
|
|
|
|
|
|
|
} |
88
|
|
|
|
|
|
|
elsif ($esc_0212) { |
89
|
0
|
|
|
|
|
0
|
$chunk =~ s/([\xa1-\xfe][\xa1-\xfe])/\x8f$1/og; |
90
|
|
|
|
|
|
|
} |
91
|
0
|
|
|
|
|
0
|
$chunk = Encode::decode( 'euc-jp', $chunk, 0 ); |
92
|
|
|
|
|
|
|
} |
93
|
|
|
|
|
|
|
elsif ( ( my $npos = index( $chunk, $trm ) ) >= 0 ) { |
94
|
0
|
|
|
|
|
0
|
$$rdst .= substr( $chunk, 0, $npos + length($trm) ); |
95
|
0
|
|
|
|
|
0
|
$$rpos += length($esc) + $npos + length($trm); |
96
|
0
|
|
|
|
|
0
|
pos($$rsrc) = $opos; |
97
|
0
|
|
|
|
|
0
|
return 1; |
98
|
|
|
|
|
|
|
} |
99
|
0
|
|
|
|
|
0
|
$$rdst .= $chunk; |
100
|
0
|
|
|
|
|
0
|
$$rpos = pos($$rsrc); |
101
|
|
|
|
|
|
|
} |
102
|
0
|
|
|
|
|
0
|
$$rpos = pos($$rsrc); |
103
|
0
|
|
|
|
|
0
|
pos($$rsrc) = $opos; |
104
|
0
|
|
|
|
|
0
|
return ''; |
105
|
|
|
|
|
|
|
} |
106
|
|
|
|
|
|
|
|
107
|
|
|
|
|
|
|
# JIS<->EUC |
108
|
|
|
|
|
|
|
my $re_scan_jis = qr{ |
109
|
|
|
|
|
|
|
(?:($RE{JIS_0212})|$RE{JIS_0208}|($RE{ISO_ASC})|($RE{JIS_KANA}))([^\e]*) |
110
|
|
|
|
|
|
|
}x; |
111
|
|
|
|
|
|
|
|
112
|
|
|
|
|
|
|
sub jis_euc { |
113
|
44
|
|
|
44
|
0
|
203
|
local ${^ENCODING}; |
114
|
44
|
|
|
|
|
114
|
my $r_str = shift; |
115
|
44
|
|
|
|
|
448
|
$$r_str =~ s($re_scan_jis) |
116
|
|
|
|
|
|
|
{ |
117
|
5060
|
|
|
|
|
17235
|
my ($esc_0212, $esc_asc, $esc_kana, $chunk) = |
118
|
|
|
|
|
|
|
($1, $2, $3, $4); |
119
|
5060
|
100
|
|
|
|
12651
|
if (!$esc_asc) { |
120
|
2530
|
|
|
|
|
5852
|
$chunk =~ tr/\x21-\x7e/\xa1-\xfe/; |
121
|
2530
|
100
|
|
|
|
7501
|
if ($esc_kana) { |
|
|
100
|
|
|
|
|
|
122
|
2
|
|
|
|
|
35
|
$chunk =~ s/([\xa1-\xdf])/\x8e$1/og; |
123
|
|
|
|
|
|
|
} |
124
|
|
|
|
|
|
|
elsif ($esc_0212) { |
125
|
426
|
|
|
|
|
6729
|
$chunk =~ s/([\xa1-\xfe][\xa1-\xfe])/\x8f$1/og; |
126
|
|
|
|
|
|
|
} |
127
|
|
|
|
|
|
|
} |
128
|
5060
|
|
|
|
|
21359
|
$chunk; |
129
|
|
|
|
|
|
|
}geox; |
130
|
44
|
|
|
|
|
511
|
my ($residue) = ( $$r_str =~ s/(\e.*)$//so ); |
131
|
44
|
|
|
|
|
214
|
return $residue; |
132
|
|
|
|
|
|
|
} |
133
|
|
|
|
|
|
|
|
134
|
|
|
|
|
|
|
sub euc_jis { |
135
|
14
|
|
|
14
|
|
7952
|
no warnings qw(uninitialized); |
|
14
|
|
|
|
|
43
|
|
|
14
|
|
|
|
|
4122
|
|
136
|
1394
|
|
|
1394
|
0
|
5706
|
local ${^ENCODING}; |
137
|
1394
|
|
|
|
|
3051
|
my $r_str = shift; |
138
|
1394
|
|
|
|
|
2762
|
my $jis0212 = shift; |
139
|
1394
|
|
|
|
|
14529
|
$$r_str =~ s{ |
140
|
|
|
|
|
|
|
((?:$RE{EUC_C})+|(?:$RE{EUC_KANA})+|(?:$RE{EUC_0212})+) |
141
|
|
|
|
|
|
|
}{ |
142
|
3467
|
|
|
|
|
9349
|
my $chunk = $1; |
143
|
|
|
|
|
|
|
my $esc = |
144
|
|
|
|
|
|
|
( $chunk =~ tr/\x8E//d ) ? $ESC{KANA} : |
145
|
|
|
|
|
|
|
( $chunk =~ tr/\x8F//d ) ? $ESC{JIS_0212} : |
146
|
3467
|
100
|
|
|
|
12826
|
$ESC{JIS_0208}; |
|
|
100
|
|
|
|
|
|
147
|
3467
|
50
|
66
|
|
|
12161
|
if ($esc eq $ESC{JIS_0212} && !$jis0212){ |
148
|
|
|
|
|
|
|
# fallback to '?' |
149
|
0
|
|
|
|
|
0
|
$chunk =~ tr/\xA1-\xFE/\x3F/; |
150
|
|
|
|
|
|
|
}else{ |
151
|
3467
|
|
|
|
|
7009
|
$chunk =~ tr/\xA1-\xFE/\x21-\x7E/; |
152
|
|
|
|
|
|
|
} |
153
|
3467
|
|
|
|
|
22827
|
$esc . $chunk . $ESC{ASC}; |
154
|
|
|
|
|
|
|
}geox; |
155
|
1394
|
|
|
|
|
4943
|
$$r_str =~ s/\Q$ESC{ASC}\E |
156
|
|
|
|
|
|
|
(\Q$ESC{KANA}\E|\Q$ESC{JIS_0212}\E|\Q$ESC{JIS_0208}\E)/$1/gox; |
157
|
1394
|
|
|
|
|
4462
|
$$r_str; |
158
|
|
|
|
|
|
|
} |
159
|
|
|
|
|
|
|
|
160
|
|
|
|
|
|
|
1; |
161
|
|
|
|
|
|
|
__END__ |