1 |
wakaba |
1.1 |
=head1 NAME |
2 |
|
|
|
3 |
|
|
Encode::ISO2022::SevenBit --- Encode and decode of 7-bit ISO/IEC 2022 |
4 |
|
|
based encodings |
5 |
|
|
|
6 |
|
|
=head1 ENCODINGS |
7 |
|
|
|
8 |
|
|
=over 4 |
9 |
|
|
|
10 |
|
|
=cut |
11 |
|
|
|
12 |
|
|
require 5.7.3; |
13 |
|
|
use strict; |
14 |
|
|
package Encode::ISO2022::SevenBit; |
15 |
|
|
use vars qw($VERSION); |
16 |
wakaba |
1.9 |
$VERSION=do{my @r=(q$Revision: 1.8 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r}; |
17 |
wakaba |
1.1 |
use base qw(Encode::Encoding); |
18 |
wakaba |
1.7 |
__PACKAGE__->Define (qw/iso-2022-7bit iso-2022-7 jis junet jis7 |
19 |
|
|
7bit-jis/); |
20 |
wakaba |
1.1 |
require Encode::ISO2022; |
21 |
|
|
|
22 |
|
|
sub encode ($$;$) { |
23 |
|
|
my ($obj, $str, $chk) = @_; |
24 |
|
|
$_[1] = '' if $chk; |
25 |
wakaba |
1.9 |
if (!defined $obj->{_encode_mapping} || $obj->{_encode_mapping}) { |
26 |
|
|
require Encode::Table; |
27 |
|
|
$str = Encode::Table::convert ($str, $obj->__encode_map, |
28 |
|
|
-autoload => defined $obj->{_encode_mapping_autoload} ? |
29 |
|
|
$obj->{_encode_mapping_autoload} : 1); |
30 |
|
|
} |
31 |
wakaba |
1.1 |
$str = &Encode::ISO2022::internal_to_iso2022 ($str, $obj->__2022_encode); |
32 |
wakaba |
1.9 |
$str; |
33 |
wakaba |
1.1 |
} |
34 |
|
|
|
35 |
|
|
sub decode ($$;$) { |
36 |
|
|
my ($obj, $str, $chk) = @_; |
37 |
|
|
$_[1] = '' if $chk; |
38 |
wakaba |
1.9 |
$str = &Encode::ISO2022::iso2022_to_internal ($str, $obj->__2022_decode); |
39 |
|
|
if (!defined $obj->{_decode_mapping} || $obj->{_decode_mapping}) { |
40 |
|
|
require Encode::Table; |
41 |
|
|
$str = Encode::Table::convert ($str, $obj->__decode_map, |
42 |
|
|
-autoload => defined $obj->{_decode_mapping_autoload} ? |
43 |
|
|
$obj->{_decode_mapping_autoload} : 1); |
44 |
|
|
} |
45 |
|
|
$str; |
46 |
wakaba |
1.1 |
} |
47 |
|
|
|
48 |
|
|
=item iso-2022-7bit |
49 |
|
|
|
50 |
|
|
ISO/IEC 2022 based 7-bit encoding using only G0 |
51 |
wakaba |
1.7 |
(Alias: junet (emacsen), jis, jis7, iso-2022-7 (emacsen), 7bit-jis (perl)) |
52 |
wakaba |
1.1 |
|
53 |
|
|
=cut |
54 |
|
|
|
55 |
|
|
sub __2022__common ($) { |
56 |
|
|
my $C = Encode::ISO2022->new_object; |
57 |
|
|
$C->{bit} = 7; |
58 |
|
|
$C->{option}->{designate_to}->{G96}->{default} = 0; |
59 |
|
|
$C->{option}->{designate_to}->{G96n}->{default} = 0; |
60 |
|
|
$C; |
61 |
|
|
} |
62 |
|
|
sub __2022_encode ($) { |
63 |
|
|
my $C = shift->__2022__common; |
64 |
wakaba |
1.4 |
$C->{GR} = undef; |
65 |
wakaba |
1.1 |
$C->{G1} = $Encode::ISO2022::CHARSET{G96}->{"\x7E"}; ## empty set |
66 |
|
|
$C; |
67 |
|
|
} |
68 |
|
|
sub __2022_decode ($) { |
69 |
|
|
my $C = shift->__2022__common; |
70 |
|
|
|
71 |
|
|
$C; |
72 |
|
|
} |
73 |
wakaba |
1.9 |
sub __encode_map ($) { |
74 |
|
|
[]; |
75 |
|
|
} |
76 |
|
|
sub __decode_map ($) { |
77 |
|
|
[]; |
78 |
|
|
} |
79 |
wakaba |
1.1 |
|
80 |
|
|
package Encode::ISO2022::SevenBit::JP; |
81 |
|
|
use vars qw/@ISA/; |
82 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit'; |
83 |
wakaba |
1.2 |
__PACKAGE__->Define (qw/iso-2022-jp junet-code japanese-iso-7bit csiso2022jp |
84 |
wakaba |
1.5 |
iso2022jp/); |
85 |
wakaba |
1.1 |
|
86 |
|
|
=item iso-2022-jp |
87 |
|
|
|
88 |
|
|
ISO/IEC 2022 based 7-bit encoding for Japanese. |
89 |
|
|
Defined by Junet no tebiki, RFC 1468 and JIS X 0208:1997 Appendix 2. |
90 |
wakaba |
1.5 |
(Alias: junet-code, japanese-iso-7bit (emacsen), csISO2022JP (IANA)) |
91 |
wakaba |
1.1 |
|
92 |
|
|
=cut |
93 |
|
|
|
94 |
|
|
sub __2022__common ($) { |
95 |
|
|
my $C = Encode::ISO2022->new_object; |
96 |
|
|
$C->{bit} = 7; |
97 |
|
|
$C->{option}->{designate_to}->{C0}->{default} = -1; |
98 |
|
|
$C->{option}->{designate_to}->{C0}->{"\x40"} = 0; |
99 |
|
|
$C->{option}->{designate_to}->{C1}->{default} = -1; |
100 |
|
|
$C->{option}->{designate_to}->{G94}->{default} = -1; |
101 |
|
|
$C->{option}->{designate_to}->{G94n}->{default} = -1; |
102 |
|
|
$C->{option}->{designate_to}->{G96}->{default} = -1; |
103 |
|
|
$C->{option}->{designate_to}->{G96n}->{default} = -1; |
104 |
|
|
$C->{option}->{designate_to}->{G94}->{"\x4A"} = 0; ## JIS X 0201 roman |
105 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x40"} = 0; ## JIS X 0208-1978 |
106 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x42"} = 0; ## JIS X 0208-1983 |
107 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x42\x40"} = 0; ## JIS X 0208-1990 |
108 |
|
|
$C->{option}->{use_revision} = 0; |
109 |
|
|
$C; |
110 |
|
|
} |
111 |
|
|
sub __2022_encode ($) { |
112 |
|
|
my $C = shift->__2022__common; |
113 |
wakaba |
1.4 |
$C->{GR} = undef; |
114 |
wakaba |
1.1 |
$C->{C1} = $Encode::ISO2022::CHARSET{C1}->{"\x7E"}; ## empty set |
115 |
|
|
$C->{G1} = $Encode::ISO2022::CHARSET{G96}->{"\x7E"}; ## empty set |
116 |
|
|
$C; |
117 |
|
|
} |
118 |
wakaba |
1.9 |
sub __encode_map ($) { |
119 |
|
|
[qw/ucs_to_ascii ucs_to_jisx0208_1983 ucs_to_jisx0208_1990 ucs_to_jisx0208_1978 ucs_to_jisx0201_latin/]; |
120 |
|
|
} |
121 |
|
|
sub __decode_map ($) { |
122 |
|
|
[qw/jisx0208_1983_to_ucs jisx0208_1990_to_ucs jisx0208_1978_to_ucs jisx0201_latin_to_ucs/]; |
123 |
|
|
} |
124 |
wakaba |
1.1 |
|
125 |
|
|
package Encode::ISO2022::SevenBit::JP1978IRV; |
126 |
|
|
use vars qw/@ISA/; |
127 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit::JP'; |
128 |
wakaba |
1.2 |
__PACKAGE__->Define (qw/iso-2022-jp-1978-irv japanese-iso-7bit-1978-irv old-jis |
129 |
wakaba |
1.6 |
x-obsoleted-iso-2022-jp jis78/); |
130 |
wakaba |
1.1 |
|
131 |
|
|
=item iso-2022-jp-1978-irv |
132 |
|
|
|
133 |
wakaba |
1.6 |
ISO/IEC 2022 based 7-bit encoding for Japanese, |
134 |
|
|
using JIS X 0208-1978. |
135 |
|
|
(Alias: japanese-iso-7bit-1978-irv (emacsen), old-jis (emacsen), |
136 |
|
|
jis78) |
137 |
wakaba |
1.1 |
|
138 |
|
|
=cut |
139 |
|
|
|
140 |
|
|
sub __2022_encode ($) { |
141 |
|
|
my $C = shift->__2022__common; |
142 |
wakaba |
1.4 |
$C->{GR} = undef; |
143 |
wakaba |
1.1 |
$C->{option}->{designate_to}->{G94}->{"\x4A"} = -1; ## JIS X 0201 roman |
144 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x42"} = -1; ## JIS X 0208-1983 |
145 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x42\x40"} = -1; ## JIS X 0208-1990 |
146 |
|
|
$C; |
147 |
|
|
} |
148 |
wakaba |
1.9 |
sub __encode_map ($) { |
149 |
|
|
[qw/ucs_to_ascii ucs_to_jisx0208_1978 ucs_to_jisx0208_1983 ucs_to_jisx0208_1990 ucs_to_jisx0201_latin/]; |
150 |
|
|
} |
151 |
|
|
sub __decode_map ($) { |
152 |
|
|
[qw/jisx0208_1983_to_ucs jisx0208_1990_to_ucs jisx0208_1978_to_ucs jisx0201_latin_to_ucs/]; |
153 |
|
|
} |
154 |
wakaba |
1.1 |
|
155 |
|
|
package Encode::ISO2022::SevenBit::JP1; |
156 |
|
|
use vars qw/@ISA/; |
157 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit::JP'; |
158 |
wakaba |
1.8 |
__PACKAGE__->Define (qw/iso-2022-jp-1 iso2022jp-1/); |
159 |
wakaba |
1.1 |
|
160 |
|
|
=item iso-2022-jp-1 |
161 |
|
|
|
162 |
|
|
ISO/IEC 2022 based 7-bit encoding for Japanese, |
163 |
wakaba |
1.8 |
defined by RFC 2237. (Alias: iso2022jp-1) |
164 |
wakaba |
1.1 |
|
165 |
|
|
=cut |
166 |
|
|
|
167 |
|
|
sub __2022__common ($) { |
168 |
|
|
my $C = shift->SUPER::__2022__common; |
169 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x44"} = 0; ## JIS X 0212-1990 |
170 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x42\x40"} = -1; ## JIS X 0208-1990 |
171 |
|
|
$C; |
172 |
|
|
} |
173 |
wakaba |
1.9 |
sub __encode_map ($) { |
174 |
|
|
[qw/ucs_to_ascii ucs_to_jisx0208_1983 ucs_to_jisx0212_1990 ucs_to_jisx0208_1978 ucs_to_jisx0201_latin/]; |
175 |
|
|
} |
176 |
|
|
sub __decode_map ($) { |
177 |
|
|
[qw/jisx0208_1983_to_ucs jisx0212_1990_to_ucs jisx0208_1978_to_ucs jisx0201_latin_to_ucs/]; |
178 |
|
|
} |
179 |
wakaba |
1.1 |
|
180 |
|
|
package Encode::ISO2022::SevenBit::JP3; |
181 |
|
|
use vars qw/@ISA/; |
182 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit::JP'; |
183 |
wakaba |
1.8 |
__PACKAGE__->Define (qw/iso-2022-jp-3 x-iso-2022-jp-3 iso2022jp-3 |
184 |
wakaba |
1.9 |
jis0213/); |
185 |
wakaba |
1.1 |
|
186 |
|
|
=item iso-2022-jp-3 |
187 |
|
|
|
188 |
|
|
ISO/IEC 2022 based 7-bit encoding for Japanese, |
189 |
|
|
defined by JIS X 0213:2000 Appendix 2. |
190 |
wakaba |
1.8 |
(Alias: x-iso-2022-jp-3, iso2022jp-3, jis0213) |
191 |
wakaba |
1.1 |
|
192 |
|
|
=cut |
193 |
|
|
|
194 |
|
|
sub __2022__common ($) { |
195 |
|
|
my $C = shift->SUPER::__2022__common; |
196 |
|
|
$C->{option}->{designate_to}->{G94}->{"\x4A"} = -1; ## JIS X 0201 roman |
197 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x40"} = -1; ## JIS X 0208-1978 |
198 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x42"} = 0; ## restricted JIS X 0213:2000 plane 1 |
199 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x42\x40"} = -1; ## JIS X 0208-1990 |
200 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x4F"} = 0; ## JIS X 0213:2000 plane 1 |
201 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x50"} = 0; ## JIS X 0213:2000 plane 2 |
202 |
|
|
$C; |
203 |
|
|
} |
204 |
wakaba |
1.9 |
sub __encode_map ($) { |
205 |
|
|
[qw/ucs_to_ascii ucs_to_jisx0213_2000_1 ucs_to_jisx0213_2000_2/]; |
206 |
|
|
} |
207 |
|
|
sub __decode_map ($) { |
208 |
|
|
[qw/jisx0213_2000_1_to_ucs jisx0212_0213_to_ucs jisx0208_1983_to_ucs jisx0208_1978_to_ucs jisx0201_latin_to_ucs/]; |
209 |
|
|
} |
210 |
|
|
|
211 |
|
|
package Encode::ISO2022::SevenBit::JP3Strict; |
212 |
|
|
use vars qw/@ISA/; |
213 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit::JP3'; |
214 |
|
|
__PACKAGE__->Define (qw/iso-2022-jp-3-strict/); |
215 |
|
|
|
216 |
|
|
=item iso-2022-jp-3-strict |
217 |
|
|
|
218 |
|
|
ISO/IEC 2022 based 7-bit encoding for Japanese. |
219 |
|
|
A subset of iso-2022-jp-3. |
220 |
|
|
See <http://www.m17n.org/m17n2000_all_but_registration/proceedings/kawabata/jisx0213.html>. |
221 |
|
|
|
222 |
|
|
=cut |
223 |
|
|
|
224 |
|
|
sub __encode_map ($) { |
225 |
|
|
[qw/ucs_to_ascii ucs_to_jisx0213_2000_1_esc_24_42 ucs_to_jisx0213_2000_1 ucs_to_jisx0213_2000_2/]; |
226 |
|
|
} |
227 |
|
|
|
228 |
|
|
package Encode::ISO2022::SevenBit::JP3Compatible; |
229 |
|
|
use vars qw/@ISA/; |
230 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit::JP3'; |
231 |
|
|
__PACKAGE__->Define (qw/iso-2022-jp-3-compatible/); |
232 |
|
|
|
233 |
|
|
=item iso-2022-jp-3-compatible |
234 |
|
|
|
235 |
|
|
ISO/IEC 2022 based 7-bit encoding for Japanese. |
236 |
|
|
See <http://www.m17n.org/m17n2000_all_but_registration/proceedings/kawabata/jisx0213.html>. |
237 |
|
|
|
238 |
|
|
=cut |
239 |
|
|
|
240 |
|
|
sub __encode_map ($) { |
241 |
|
|
[qw/ucs_to_ascii ucs_to_jisx0208_1983 ucs_to_jisx0213_2000_1 ucs_to_jisx0213_2000_2/]; |
242 |
|
|
} |
243 |
wakaba |
1.1 |
|
244 |
|
|
package Encode::ISO2022::SevenBit::JP3Plane1; |
245 |
|
|
use vars qw/@ISA/; |
246 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit::JP3'; |
247 |
|
|
__PACKAGE__->Define (qw/iso-2022-jp-3-plane1/); |
248 |
|
|
|
249 |
|
|
=item iso-2022-jp-3-plane1 |
250 |
|
|
|
251 |
|
|
ISO/IEC 2022 based 7-bit encoding for Japanese, |
252 |
|
|
defined by JIS X 0213:2000 Appendix 2. |
253 |
|
|
|
254 |
|
|
=cut |
255 |
|
|
|
256 |
|
|
sub __2022__common ($) { |
257 |
|
|
my $C = shift->SUPER::__2022__common; |
258 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x50"} = -1; ## JIS X 0213:2000 plane 2 |
259 |
|
|
$C; |
260 |
|
|
} |
261 |
wakaba |
1.9 |
sub __encode_map ($) { |
262 |
|
|
[qw/ucs_to_ascii ucs_to_jisx0213_2000_1/]; |
263 |
|
|
} |
264 |
|
|
sub __decode_map ($) { |
265 |
|
|
[qw/jisx0213_2000_1_to_ucs jisx0208_1983_to_ucs/]; |
266 |
|
|
} |
267 |
wakaba |
1.1 |
|
268 |
|
|
package Encode::ISO2022::SevenBit::SS2; |
269 |
|
|
use vars qw/@ISA/; |
270 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit'; |
271 |
wakaba |
1.3 |
__PACKAGE__->Define (qw/iso-2022-7bit-ss2 x-iso-2022-jp-2 iso-2022-ss2-7 |
272 |
|
|
jis_encoding csjisencoding/); |
273 |
wakaba |
1.1 |
|
274 |
|
|
=item iso-2022-7bit-ss2 |
275 |
|
|
|
276 |
|
|
ISO/IEC 2022 based 7-bit encoding using SS2 for 96-charset |
277 |
wakaba |
1.3 |
(Alias: x-iso-2022-jp-2, iso-2022-ss2-7 (emacsen)) |
278 |
wakaba |
1.1 |
|
279 |
|
|
=cut |
280 |
|
|
|
281 |
|
|
sub __2022__common ($) { |
282 |
|
|
my $C = shift->SUPER::__2022__common; |
283 |
|
|
$C->{option}->{designate_to}->{G96}->{default} = 2; |
284 |
|
|
$C->{option}->{designate_to}->{G96n}->{default} = 2; |
285 |
|
|
$C->{option}->{Ginvoke_by_single_shift}->[2] = 1; |
286 |
|
|
$C; |
287 |
|
|
} |
288 |
|
|
|
289 |
|
|
package Encode::ISO2022::SevenBit::JP2; |
290 |
|
|
use vars qw/@ISA/; |
291 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit::SS2'; |
292 |
wakaba |
1.8 |
__PACKAGE__->Define (qw/iso-2022-jp-2 csiso2022jp2 iso2022jp-2/); |
293 |
wakaba |
1.1 |
|
294 |
|
|
=item iso-2022-jp-2 |
295 |
|
|
|
296 |
|
|
ISO/IEC 2022 based 7-bit multilingual encoding, defined by |
297 |
wakaba |
1.8 |
RFC 1554. A subset of iso-2022-7bit-ss2. (Alias: iso2022jp-2, |
298 |
|
|
csISO2022JP2 (IANA)) |
299 |
wakaba |
1.1 |
|
300 |
|
|
=cut |
301 |
|
|
|
302 |
|
|
sub __2022__common ($) { |
303 |
|
|
my $C = shift->SUPER::__2022__common; |
304 |
|
|
$C->{option}->{designate_to}->{C0}->{default} = -1; |
305 |
|
|
$C->{option}->{designate_to}->{C0}->{"\x40"} = 0; |
306 |
|
|
$C->{option}->{designate_to}->{C1}->{default} = -1; |
307 |
|
|
$C->{C1} = $Encode::ISO2022::CHARSET{C1}->{"\x47"}; ## Minimum C1 |
308 |
wakaba |
1.9 |
$C->{option}->{designate_to}->{C1}->{"\x47"} = -1; |
309 |
wakaba |
1.1 |
$C->{option}->{designate_to}->{G94}->{default} = -1; |
310 |
|
|
$C->{option}->{designate_to}->{G94n}->{default} = -1; |
311 |
|
|
$C->{option}->{designate_to}->{G96}->{default} = -1; |
312 |
|
|
$C->{option}->{designate_to}->{G96n}->{default} = -1; |
313 |
|
|
$C->{option}->{designate_to}->{G94}->{"\x4A"} = 0; ## JIS X 0201 roman |
314 |
wakaba |
1.9 |
for ("\x40", "\x41", "\x42", "\x43", "\x44") { |
315 |
wakaba |
1.1 |
$C->{option}->{designate_to}->{G94n}->{ $_ } = 0; |
316 |
|
|
} |
317 |
|
|
for ("\x41", "\x46") { |
318 |
|
|
$C->{option}->{designate_to}->{G96}->{ $_ } = 2; |
319 |
|
|
} |
320 |
|
|
$C; |
321 |
|
|
} |
322 |
|
|
sub __2022_encode ($) { |
323 |
|
|
my $C = shift->__2022__common; |
324 |
wakaba |
1.4 |
$C->{GR} = undef; |
325 |
wakaba |
1.1 |
$C->{C1} = $Encode::ISO2022::CHARSET{C1}->{"\x7E"}; ## empty set |
326 |
|
|
$C->{G1} = $Encode::ISO2022::CHARSET{G96}->{"\x7E"}; ## empty set |
327 |
|
|
$C; |
328 |
|
|
} |
329 |
wakaba |
1.9 |
sub __encode_map ($) { |
330 |
|
|
[qw/ucs_to_ascii ucs_to_jisx0208_1983 ucs_to_jisx0212_1990 ucs_to_gb2312_1980 ucs_to_ksx1001_1992 ucs_to_jisx0208_1978 ucs_to_isoiec8859_1 ucs_to_isoiec8859_7 ucs_to_jisx0201_latin/]; |
331 |
|
|
} |
332 |
|
|
sub __decode_map ($) { |
333 |
|
|
[qw/jisx0208_1983_to_ucs jisx0208_1978_to_ucs jisx0201_latin_to_ucs gb2312_1980_to_ucs ksx1001_1992_to_ucs isoiec8859_7_to_ucs/]; |
334 |
|
|
} |
335 |
wakaba |
1.1 |
|
336 |
|
|
package Encode::ISO2022::SevenBit::Lock; |
337 |
|
|
use vars qw/@ISA/; |
338 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit'; |
339 |
|
|
__PACKAGE__->Define (qw/iso-2022-7bit-lock/); |
340 |
|
|
|
341 |
|
|
=item iso-2022-7bit-lock |
342 |
|
|
|
343 |
|
|
ISO/IEC 2022 based 7-bit encoding using G1 and locking-shift for 96-charset |
344 |
|
|
|
345 |
|
|
=cut |
346 |
|
|
|
347 |
|
|
sub __2022__common ($) { |
348 |
|
|
my $C = shift->SUPER::__2022__common; |
349 |
|
|
$C->{option}->{designate_to}->{G96}->{default} = 1; |
350 |
|
|
$C->{option}->{designate_to}->{G96n}->{default} = 1; |
351 |
|
|
$C; |
352 |
|
|
} |
353 |
|
|
|
354 |
wakaba |
1.3 |
package Encode::ISO2022::SevenBit::INT; |
355 |
|
|
use vars qw/@ISA/; |
356 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit'; |
357 |
|
|
__PACKAGE__->Define (qw/iso-2022-int iso-2022-int-* iso-2022-int-2/); |
358 |
|
|
|
359 |
|
|
=item iso-2022-int |
360 |
|
|
|
361 |
|
|
ISO/IEC 2022 based 7-bit encoding using G1 and locking-shift for |
362 |
|
|
KS X 1001 and 96-charset. See draft-ohta-text-encoding. |
363 |
|
|
(Alias: iso-2022-int-*) |
364 |
|
|
|
365 |
|
|
=cut |
366 |
|
|
|
367 |
|
|
sub __2022__common ($) { |
368 |
|
|
my $C = shift->SUPER::__2022__common; |
369 |
|
|
$C->{option}->{designate_to}->{G96}->{default} = 1; |
370 |
|
|
$C->{option}->{designate_to}->{G96n}->{default} = 1; |
371 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x43"} = 1; ## KS X 1001 |
372 |
|
|
$C; |
373 |
|
|
} |
374 |
|
|
sub __2022_encode ($) { |
375 |
|
|
my $C = shift->__2022__common; |
376 |
wakaba |
1.4 |
$C->{GR} = undef; |
377 |
wakaba |
1.3 |
$C->{C1} = $Encode::ISO2022::CHARSET{C1}->{"\x7E"}; ## empty set |
378 |
|
|
$C->{G1} = $Encode::ISO2022::CHARSET{G96}->{"\x7E"}; ## empty set |
379 |
|
|
$C; |
380 |
|
|
} |
381 |
|
|
sub __2022_decode ($) { |
382 |
|
|
my $C = shift->__2022__common; |
383 |
|
|
$C->{G1} = $Encode::ISO2022::CHARSET{G94n}->{"\x43"}; ## KS X 1001 |
384 |
|
|
$C; |
385 |
|
|
} |
386 |
|
|
|
387 |
wakaba |
1.1 |
package Encode::ISO2022::SevenBit::KR; |
388 |
|
|
use vars qw/@ISA/; |
389 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit'; |
390 |
wakaba |
1.2 |
__PACKAGE__->Define (qw/iso-2022-kr korean-iso-7bit csiso2022kr cp50225 kr2022/); |
391 |
wakaba |
1.1 |
|
392 |
|
|
=item iso-2022-kr |
393 |
|
|
|
394 |
|
|
An ISO/IEC 2022 based 7-bit encoding for Korean, |
395 |
wakaba |
1.2 |
defined by RFC 1557 (Alias: korean-iso-7bit (emacsen), |
396 |
|
|
csISO2022KR (IANA), CP50225 (M$), KR2022) |
397 |
wakaba |
1.1 |
|
398 |
|
|
=cut |
399 |
|
|
|
400 |
|
|
sub __2022__common ($) { |
401 |
|
|
my $C = shift->SUPER::__2022__common; |
402 |
|
|
$C->{option}->{designate_to}->{C0}->{default} = -1; |
403 |
|
|
$C->{option}->{designate_to}->{C0}->{"\x40"} = 0; |
404 |
|
|
$C->{option}->{designate_to}->{C1}->{default} = -1; |
405 |
|
|
$C->{option}->{designate_to}->{G94}->{default} = -1; |
406 |
|
|
$C->{option}->{designate_to}->{G94n}->{default} = -1; |
407 |
|
|
$C->{option}->{designate_to}->{G96}->{default} = -1; |
408 |
|
|
$C->{option}->{designate_to}->{G96n}->{default} = -1; |
409 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x43"} = 1; ## KS X 1001 |
410 |
|
|
$C; |
411 |
|
|
} |
412 |
|
|
sub __2022_encode ($) { |
413 |
|
|
my $C = shift->__2022__common; |
414 |
wakaba |
1.4 |
$C->{GR} = undef; |
415 |
wakaba |
1.1 |
$C->{C1} = $Encode::ISO2022::CHARSET{C1}->{"\x7E"}; ## empty set |
416 |
|
|
$C->{G1} = $Encode::ISO2022::CHARSET{G96}->{"\x7E"}; ## empty set |
417 |
|
|
$C; |
418 |
|
|
} |
419 |
|
|
sub __2022_decode ($) { |
420 |
|
|
my $C = shift->__2022__common; |
421 |
|
|
$C->{G1} = $Encode::ISO2022::CHARSET{G94n}->{"\x43"}; ## KS X 1001 |
422 |
|
|
$C; |
423 |
|
|
} |
424 |
wakaba |
1.9 |
sub __encode_map ($) { |
425 |
|
|
[qw/ucs_to_ascii ksx1001_1992_to_ucs/]; |
426 |
|
|
} |
427 |
|
|
sub __decode_map ($) { |
428 |
|
|
[qw/ksx1001_1992_to_ucs/]; |
429 |
|
|
} |
430 |
wakaba |
1.1 |
|
431 |
|
|
package Encode::ISO2022::SevenBit::INT1; |
432 |
|
|
use vars qw/@ISA/; |
433 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit'; |
434 |
|
|
__PACKAGE__->Define (qw/iso-2022-int-1/); |
435 |
|
|
|
436 |
|
|
=item iso-2022-int-1 |
437 |
|
|
|
438 |
|
|
An ISO/IEC 2022 based 7-bit multilingual encoding, |
439 |
|
|
defined by draft-ohta-text-encoding |
440 |
|
|
|
441 |
|
|
=cut |
442 |
|
|
|
443 |
|
|
sub __2022__common ($) { |
444 |
|
|
my $C = shift->SUPER::__2022__common; |
445 |
|
|
$C->{option}->{designate_to}->{C0}->{default} = -1; |
446 |
|
|
$C->{option}->{designate_to}->{C0}->{"\x40"} = 0; |
447 |
|
|
$C->{option}->{designate_to}->{C1}->{default} = -1; |
448 |
|
|
$C->{option}->{designate_to}->{G94}->{default} = -1; |
449 |
|
|
$C->{option}->{designate_to}->{G94n}->{default} = -1; |
450 |
|
|
$C->{option}->{designate_to}->{G96}->{default} = -1; |
451 |
|
|
$C->{option}->{designate_to}->{G96n}->{default} = -1; |
452 |
|
|
$C->{option}->{designate_to}->{G94}->{"\x4A"} = 0; ## JIS X 0201 roman |
453 |
wakaba |
1.9 |
for ("\x40", "\x41", "\x42", "\x44") { |
454 |
wakaba |
1.1 |
$C->{option}->{designate_to}->{G94n}->{ $_ } = 0; |
455 |
|
|
} |
456 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x43"} = 1; ## KS X 1001 |
457 |
|
|
for ("\x41", "\x46") { ## ISO/IEC 8859-1,7 |
458 |
|
|
$C->{option}->{designate_to}->{G96}->{ $_ } = 1; |
459 |
|
|
} |
460 |
|
|
$C; |
461 |
|
|
} |
462 |
|
|
sub __2022_encode ($) { |
463 |
|
|
my $C = shift->__2022__common; |
464 |
wakaba |
1.4 |
$C->{GR} = undef; |
465 |
wakaba |
1.1 |
$C->{C1} = $Encode::ISO2022::CHARSET{C1}->{"\x7E"}; ## empty set |
466 |
|
|
$C->{G1} = $Encode::ISO2022::CHARSET{G96}->{"\x7E"}; ## empty set |
467 |
|
|
$C; |
468 |
|
|
} |
469 |
|
|
sub __2022_decode ($) { |
470 |
|
|
my $C = shift->__2022__common; |
471 |
|
|
$C->{G1} = $Encode::ISO2022::CHARSET{G94n}->{"\x43"}; ## KS X 1001 |
472 |
|
|
$C; |
473 |
|
|
} |
474 |
wakaba |
1.9 |
sub __encode_map ($) { |
475 |
|
|
[qw/ucs_to_ascii ucs_to_isoiec8859_1 ucs_to_jisx0208_1983 ucs_to_jisx0212_1990 ucs_to_gb2312_1980 ucs_to_ksx1001_1992 ucs_to_isoiec8859_7 ucs_to_jisx0208_1978 ucs_to_jisx0201_latin/]; |
476 |
|
|
} |
477 |
|
|
sub __decode_map ($) { |
478 |
|
|
[qw/jisx0208_1983_to_ucs jisx0208_1978_to_ucs jisx0201_latin_to_ucs gb2312_1980_to_ucs ksx1001_1992_to_ucs isoiec8859_7_to_ucs/]; |
479 |
|
|
} |
480 |
wakaba |
1.1 |
|
481 |
|
|
package Encode::ISO2022::SevenBit::LockSS2; |
482 |
|
|
use vars qw/@ISA/; |
483 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit'; |
484 |
|
|
__PACKAGE__->Define (qw/iso-2022-7bit-lock-ss2 iso-2022-cjk/); |
485 |
|
|
|
486 |
|
|
=item iso-2022-7bit-lock-ss2 |
487 |
|
|
|
488 |
|
|
An ISO/IEC 2022 based 7-bit encoding. Mixture of ISO-2022-JP, |
489 |
|
|
ISO-2022-KR, ISO-2022-CN. (Alias: iso-2022-cjk (emacsen)) |
490 |
|
|
|
491 |
|
|
=cut |
492 |
|
|
|
493 |
|
|
sub __2022__common ($) { |
494 |
|
|
my $C = shift->SUPER::__2022__common; |
495 |
|
|
$C->{option}->{designate_to}->{G96}->{default} = 2; |
496 |
|
|
$C->{option}->{designate_to}->{G96n}->{default} = 2; |
497 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x41"} = 1; ## GB 2312 |
498 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x43"} = 1; ## KS X 1001 |
499 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x45"} = 1; ## ISO-IR 165 |
500 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x47"} = 1; ## CNS 11643 plane 1 |
501 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x48"} = 2; ## CNS 11643 plane 2 |
502 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x49"} = 3; ## CNS 11643 plane 3 |
503 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x4A"} = 3; ## CNS 11643 plane 4 |
504 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x4B"} = 3; ## CNS 11643 plane 5 |
505 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x4C"} = 3; ## CNS 11643 plane 6 |
506 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x4D"} = 3; ## CNS 11643 plane 7 |
507 |
wakaba |
1.2 |
$C->{option}->{designate_to}->{G94n}->{P0_0} = 1; ## GB 12345 |
508 |
|
|
$C->{option}->{designate_to}->{G94n}->{P0_1} = 2; ## GB 7589 |
509 |
|
|
$C->{option}->{designate_to}->{G94n}->{P0_2} = 2; ## GB 13131 |
510 |
|
|
$C->{option}->{designate_to}->{G94n}->{P0_3} = 3; ## GB 7590 |
511 |
|
|
$C->{option}->{designate_to}->{G94n}->{P0_4} = 3; ## GB 13132 |
512 |
wakaba |
1.1 |
$C; |
513 |
|
|
} |
514 |
|
|
sub __2022_decode ($) { |
515 |
|
|
my $C = shift->__2022__common; |
516 |
|
|
$C->{G1} = $Encode::ISO2022::CHARSET{G94n}->{"\x43"}; ## KS X 1001 |
517 |
|
|
$C; |
518 |
|
|
} |
519 |
|
|
|
520 |
|
|
package Encode::ISO2022::SevenBit::CN; |
521 |
|
|
use vars qw/@ISA/; |
522 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit'; |
523 |
wakaba |
1.2 |
__PACKAGE__->Define (qw/iso-2022-cn chinese-iso-7bit iso2022cn-cns iso2022cn-gb/); |
524 |
wakaba |
1.1 |
|
525 |
|
|
=item iso-2022-cn |
526 |
|
|
|
527 |
|
|
An ISO/IEC 2022 based 7-bit encoding for Chinese, |
528 |
|
|
defined by RFC 1922 (Alias: chinese-iso-7bit (emacsen)) |
529 |
|
|
|
530 |
|
|
=cut |
531 |
|
|
|
532 |
|
|
sub __2022__common ($) { |
533 |
|
|
my $C = shift->SUPER::__2022__common; |
534 |
|
|
$C->{option}->{designate_to}->{C0}->{default} = -1; |
535 |
|
|
$C->{option}->{designate_to}->{C0}->{"\x40"} = 0; |
536 |
|
|
$C->{option}->{designate_to}->{C1}->{default} = -1; |
537 |
|
|
$C->{option}->{designate_to}->{G94}->{default} = -1; |
538 |
|
|
$C->{option}->{designate_to}->{G94n}->{default} = -1; |
539 |
|
|
$C->{option}->{designate_to}->{G96}->{default} = -1; |
540 |
|
|
$C->{option}->{designate_to}->{G96n}->{default} = -1; |
541 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x41"} = 1; ## GB 2312 |
542 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x47"} = 1; ## CNS 11643 plane 1 |
543 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x48"} = 2; ## CNS 11643 plane 2 |
544 |
|
|
$C; |
545 |
|
|
} |
546 |
|
|
sub __2022_encode ($) { |
547 |
|
|
my $C = shift->__2022__common; |
548 |
wakaba |
1.4 |
$C->{GR} = undef; |
549 |
wakaba |
1.1 |
$C->{C1} = $Encode::ISO2022::CHARSET{C1}->{"\x7E"}; ## empty set |
550 |
|
|
$C->{G1} = $Encode::ISO2022::CHARSET{G96}->{"\x7E"}; ## empty set |
551 |
|
|
$C; |
552 |
|
|
} |
553 |
|
|
|
554 |
|
|
package Encode::ISO2022::SevenBit::CNExt; |
555 |
|
|
use vars qw/@ISA/; |
556 |
|
|
push @ISA, 'Encode::ISO2022::SevenBit::CN'; |
557 |
|
|
__PACKAGE__->Define (qw/iso-2022-cn-ext/); |
558 |
|
|
|
559 |
|
|
=item iso-2022-cn-ext |
560 |
|
|
|
561 |
|
|
An ISO/IEC 2022 based 7-bit encoding for Chinese, |
562 |
|
|
defined by RFC 1922 |
563 |
|
|
|
564 |
|
|
=cut |
565 |
|
|
|
566 |
|
|
sub __2022__common ($) { |
567 |
|
|
my $C = shift->SUPER::__2022__common; |
568 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x45"} = 1; ## ISO-IR 165 |
569 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x49"} = 3; ## CNS 11643 plane 3 |
570 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x4A"} = 3; ## CNS 11643 plane 4 |
571 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x4B"} = 3; ## CNS 11643 plane 5 |
572 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x4C"} = 3; ## CNS 11643 plane 6 |
573 |
|
|
$C->{option}->{designate_to}->{G94n}->{"\x4D"} = 3; ## CNS 11643 plane 7 |
574 |
wakaba |
1.2 |
$C->{option}->{designate_to}->{G94n}->{P0_0} = 1; ## GB 12345 |
575 |
|
|
$C->{option}->{designate_to}->{G94n}->{P0_1} = 2; ## GB 7589 |
576 |
|
|
$C->{option}->{designate_to}->{G94n}->{P0_2} = 2; ## GB 13131 |
577 |
|
|
$C->{option}->{designate_to}->{G94n}->{P0_3} = 3; ## GB 7590 |
578 |
|
|
$C->{option}->{designate_to}->{G94n}->{P0_4} = 3; ## GB 13132 |
579 |
|
|
$C->{option}->{designate_to}->{G94n}->{P1_0} = 3; ## CNS 11643 plane 8 |
580 |
|
|
$C->{option}->{designate_to}->{G94n}->{P1_1} = 3; ## CNS 11643 plane 9 |
581 |
|
|
$C->{option}->{designate_to}->{G94n}->{P1_2} = 3; ## CNS 11643 plane 10 |
582 |
|
|
$C->{option}->{designate_to}->{G94n}->{P1_3} = 3; ## CNS 11643 plane 11 |
583 |
|
|
$C->{option}->{designate_to}->{G94n}->{P1_4} = 3; ## CNS 11643 plane 12 |
584 |
|
|
$C->{option}->{designate_to}->{G94n}->{P1_5} = 3; ## CNS 11643 plane 13 |
585 |
|
|
$C->{option}->{designate_to}->{G94n}->{P1_6} = 3; ## CNS 11643 plane 14 |
586 |
|
|
$C->{option}->{designate_to}->{G94n}->{P2_0} = 3; ## CNS 11643 plane 15 |
587 |
|
|
$C->{option}->{designate_to}->{G94n}->{P2_1} = 3; ## CNS 11643 plane 16 |
588 |
wakaba |
1.1 |
$C; |
589 |
|
|
} |
590 |
|
|
|
591 |
|
|
1; |
592 |
|
|
__END__ |
593 |
|
|
|
594 |
|
|
=back |
595 |
|
|
|
596 |
wakaba |
1.9 |
=head1 AUTHORS |
597 |
|
|
|
598 |
|
|
Wakaba <w@suika.fam.cx> |
599 |
|
|
|
600 |
|
|
Nanashi-san |
601 |
|
|
|
602 |
wakaba |
1.1 |
=head1 LICENSE |
603 |
|
|
|
604 |
wakaba |
1.9 |
Copyright 2002 AUTHORS |
605 |
wakaba |
1.1 |
|
606 |
|
|
This library is free software; you can redistribute it |
607 |
|
|
and/or modify it under the same terms as Perl itself. |
608 |
|
|
|
609 |
|
|
=cut |
610 |
|
|
|
611 |
wakaba |
1.9 |
# $Date: 2002/10/04 23:58:04 $ |
612 |
wakaba |
1.1 |
### SevenBit.pm ends here |