1 |
wakaba |
1.1 |
## This file is auto-generated (at 2007-07-19T12:15:40Z). |
2 |
|
|
## Do not edit by hand! |
3 |
|
|
use strict; |
4 |
|
|
|
5 |
|
|
package Char::Class::JISX4051; |
6 |
|
|
our $VERSION = '1.01'; |
7 |
|
|
|
8 |
|
|
use Exporter; |
9 |
|
|
use vars qw(@EXPORT_OK @ISA $VERSION); |
10 |
|
|
@ISA = qw(Exporter); |
11 |
|
|
|
12 |
|
|
=head1 NAME |
13 |
|
|
|
14 |
|
|
Char::Class::JISX4051 - Regular Expression Character Classes - C<JISX4051> |
15 |
|
|
|
16 |
|
|
=head1 DESCRIPTION |
17 |
|
|
|
18 |
|
|
JIS X 4051 character classes represented in UCS, |
19 |
|
|
defined by JIS X 4051 Appendix |
20 |
|
|
|
21 |
|
|
|
22 |
|
|
=cut |
23 |
|
|
|
24 |
|
|
sub import ($;@) { |
25 |
|
|
my ($self, @sub) = (shift, @_); |
26 |
|
|
for (@sub) { |
27 |
|
|
no strict 'refs'; |
28 |
|
|
*{'main::'.$_} = \&{$_}; |
29 |
|
|
} |
30 |
|
|
$Exporter::ExportLevel = 1; |
31 |
|
|
$self->SUPER::import (@_); |
32 |
|
|
$Exporter::ExportLevel = 0; |
33 |
|
|
} |
34 |
|
|
|
35 |
|
|
@EXPORT_OK = qw(InJISX4051AlphabeticExceptSpace InJISX4051AlphabeticSpace InJISX4051CharacterInDigits InJISX4051CharacterInUnitSymbol InJISX4051CloseBracket InJISX4051FullStop InJISX4051Hiragana InJISX4051JapaneseSpace InJISX4051LineStartKinsokuJapaneseCharacter InJISX4051MiddleDot InJISX4051MiscJapaneseCharacter InJISX4051OpenBracket InJISX4051PostShortenedSymbol InJISX4051PreShortenedSymbol InJISX4051SeparationSymbol InJISX4051UnseparatableCharacter InJISX4051WarichuuCloseBracket InJISX4051WarichuuOpenBracket); |
36 |
|
|
|
37 |
|
|
sub InJISX4051AlphabeticExceptSpace { |
38 |
|
|
<<EOH; |
39 |
|
|
0021 007E |
40 |
|
|
00A1 00FF |
41 |
|
|
0100 017F |
42 |
|
|
0250 02A8 |
43 |
|
|
0374 0375 |
44 |
|
|
037A |
45 |
|
|
037E |
46 |
|
|
0384 038A |
47 |
|
|
038C |
48 |
|
|
038E 03A1 |
49 |
|
|
03A3 03CE |
50 |
|
|
0401 040C |
51 |
|
|
040E 044F |
52 |
|
|
0451 045C |
53 |
|
|
045E 047F |
54 |
|
|
0480 0486 |
55 |
|
|
0490 04C4 |
56 |
|
|
04C7 04C8 |
57 |
|
|
04CB 04CC |
58 |
|
|
04D0 04EB |
59 |
|
|
04EE 04F5 |
60 |
|
|
04F8 04F9 |
61 |
|
|
2000 202E |
62 |
|
|
2030 2046 |
63 |
|
|
2070 |
64 |
|
|
2074 208E |
65 |
|
|
20A0 20AA |
66 |
|
|
2100 2138 |
67 |
|
|
2153 2182 |
68 |
|
|
2190 21EA |
69 |
|
|
2200 227F |
70 |
|
|
2280 22F1 |
71 |
|
|
2312 |
72 |
|
|
2460 24EA |
73 |
|
|
2500 257F |
74 |
|
|
2580 2595 |
75 |
|
|
25A0 25EF |
76 |
|
|
2600 2613 |
77 |
|
|
261A 266F |
78 |
|
|
2701 2704 |
79 |
|
|
2706 2709 |
80 |
|
|
270C 2727 |
81 |
|
|
2729 274B |
82 |
|
|
274D |
83 |
|
|
274F 2752 |
84 |
|
|
2756 |
85 |
|
|
2758 275E |
86 |
|
|
2761 2767 |
87 |
|
|
2776 2794 |
88 |
|
|
2798 27AF |
89 |
|
|
27B1 27BE |
90 |
|
|
EOH |
91 |
|
|
} |
92 |
|
|
|
93 |
|
|
sub InJISX4051AlphabeticSpace { |
94 |
|
|
<<EOH; |
95 |
|
|
0020 |
96 |
|
|
EOH |
97 |
|
|
} |
98 |
|
|
|
99 |
|
|
sub InJISX4051CharacterInDigits { |
100 |
|
|
<<EOH; |
101 |
|
|
0030 0039 |
102 |
|
|
002C |
103 |
|
|
002E |
104 |
|
|
EOH |
105 |
|
|
} |
106 |
|
|
|
107 |
|
|
sub InJISX4051CharacterInUnitSymbol { |
108 |
|
|
<<EOH; |
109 |
|
|
0041 005A |
110 |
|
|
0061 007A |
111 |
|
|
2126 |
112 |
|
|
EOH |
113 |
|
|
} |
114 |
|
|
|
115 |
|
|
sub InJISX4051CloseBracket { |
116 |
|
|
<<EOH; |
117 |
|
|
0029 |
118 |
|
|
002C |
119 |
|
|
005D |
120 |
|
|
007D |
121 |
|
|
2019 |
122 |
|
|
201A |
123 |
|
|
201D |
124 |
|
|
201E |
125 |
|
|
3001 |
126 |
|
|
3009 |
127 |
|
|
300B |
128 |
|
|
300D |
129 |
|
|
300F |
130 |
|
|
3011 |
131 |
|
|
3015 |
132 |
|
|
3017 |
133 |
|
|
3019 |
134 |
|
|
301B |
135 |
|
|
301E |
136 |
|
|
301F |
137 |
|
|
EOH |
138 |
|
|
} |
139 |
|
|
|
140 |
|
|
sub InJISX4051FullStop { |
141 |
|
|
<<EOH; |
142 |
|
|
002E |
143 |
|
|
3002 |
144 |
|
|
EOH |
145 |
|
|
} |
146 |
|
|
|
147 |
|
|
sub InJISX4051Hiragana { |
148 |
|
|
<<EOH; |
149 |
|
|
3042 |
150 |
|
|
3044 |
151 |
|
|
3046 |
152 |
|
|
3048 |
153 |
|
|
304A 3062 |
154 |
|
|
3064 3082 |
155 |
|
|
3084 |
156 |
|
|
3086 |
157 |
|
|
3088 308D |
158 |
|
|
308F 3094 |
159 |
|
|
EOH |
160 |
|
|
} |
161 |
|
|
|
162 |
|
|
sub InJISX4051JapaneseSpace { |
163 |
|
|
<<EOH; |
164 |
|
|
3000 |
165 |
|
|
EOH |
166 |
|
|
} |
167 |
|
|
|
168 |
|
|
sub InJISX4051LineStartKinsokuJapaneseCharacter { |
169 |
|
|
<<EOH; |
170 |
|
|
203C |
171 |
|
|
2044 |
172 |
|
|
301C |
173 |
|
|
3041 |
174 |
|
|
3043 |
175 |
|
|
3045 |
176 |
|
|
3047 |
177 |
|
|
3049 |
178 |
|
|
3063 |
179 |
|
|
3083 |
180 |
|
|
3085 |
181 |
|
|
3087 |
182 |
|
|
308E |
183 |
|
|
309D |
184 |
|
|
309E |
185 |
|
|
30A1 |
186 |
|
|
30A3 |
187 |
|
|
30A5 |
188 |
|
|
30A7 |
189 |
|
|
30A9 |
190 |
|
|
30C3 |
191 |
|
|
30E3 |
192 |
|
|
30E5 |
193 |
|
|
30E7 |
194 |
|
|
30EE |
195 |
|
|
30F5 |
196 |
|
|
30F6 |
197 |
|
|
30FC |
198 |
|
|
30FD |
199 |
|
|
30FE |
200 |
|
|
EOH |
201 |
|
|
} |
202 |
|
|
|
203 |
|
|
sub InJISX4051MiddleDot { |
204 |
|
|
<<EOH; |
205 |
|
|
003A |
206 |
|
|
003B |
207 |
|
|
00FB |
208 |
|
|
EOH |
209 |
|
|
} |
210 |
|
|
|
211 |
|
|
sub InJISX4051MiscJapaneseCharacter { |
212 |
|
|
<<EOH; |
213 |
|
|
002B |
214 |
|
|
002D |
215 |
|
|
003C |
216 |
|
|
003D |
217 |
|
|
003E |
218 |
|
|
00A7 |
219 |
|
|
00A9 |
220 |
|
|
00AE |
221 |
|
|
00B1 |
222 |
|
|
00B6 |
223 |
|
|
00D7 |
224 |
|
|
00F7 |
225 |
|
|
203B |
226 |
|
|
2160 217F |
227 |
|
|
2190 21EA |
228 |
|
|
2460 24EA |
229 |
|
|
2500 257F |
230 |
|
|
2580 2595 |
231 |
|
|
25A0 25EF |
232 |
|
|
2600 2613 |
233 |
|
|
261A 266E |
234 |
|
|
2701 2704 |
235 |
|
|
2706 2709 |
236 |
|
|
270C 2727 |
237 |
|
|
2729 274B |
238 |
|
|
274D |
239 |
|
|
274F 2752 |
240 |
|
|
2756 |
241 |
|
|
2758 275E |
242 |
|
|
2761 2767 |
243 |
|
|
2776 2794 |
244 |
|
|
2798 27AF |
245 |
|
|
27B1 27BE |
246 |
|
|
3003 |
247 |
|
|
3004 |
248 |
|
|
3006 |
249 |
|
|
3007 |
250 |
|
|
3012 |
251 |
|
|
3013 |
252 |
|
|
3020 |
253 |
|
|
3036 |
254 |
|
|
30A2 |
255 |
|
|
30A4 |
256 |
|
|
30A6 |
257 |
|
|
30A8 |
258 |
|
|
30AA 30C2 |
259 |
|
|
30C4 30E2 |
260 |
|
|
30E4 |
261 |
|
|
30E6 |
262 |
|
|
30E8 30ED |
263 |
|
|
30EF 30F4 |
264 |
|
|
30F7 30FA |
265 |
|
|
3220 3243 |
266 |
|
|
3280 32B0 |
267 |
|
|
32D0 32FE |
268 |
|
|
3300 3357 |
269 |
|
|
3371 3376 |
270 |
|
|
3380 33DD |
271 |
|
|
4E00 9FA5 |
272 |
|
|
EOH |
273 |
|
|
} |
274 |
|
|
|
275 |
|
|
sub InJISX4051OpenBracket { |
276 |
|
|
<<EOH; |
277 |
|
|
0028 |
278 |
|
|
005B |
279 |
|
|
007B |
280 |
|
|
2018 |
281 |
|
|
201B |
282 |
|
|
201C |
283 |
|
|
201F |
284 |
|
|
3008 |
285 |
|
|
300A |
286 |
|
|
300C |
287 |
|
|
300E |
288 |
|
|
3010 |
289 |
|
|
3014 |
290 |
|
|
3016 |
291 |
|
|
3018 |
292 |
|
|
301A |
293 |
|
|
301D |
294 |
|
|
EOH |
295 |
|
|
} |
296 |
|
|
|
297 |
|
|
sub InJISX4051PostShortenedSymbol { |
298 |
|
|
<<EOH; |
299 |
|
|
0025 |
300 |
|
|
00A2 |
301 |
|
|
00B0 |
302 |
|
|
2030 2033 |
303 |
|
|
EOH |
304 |
|
|
} |
305 |
|
|
|
306 |
|
|
sub InJISX4051PreShortenedSymbol { |
307 |
|
|
<<EOH; |
308 |
|
|
0024 |
309 |
|
|
00A3 |
310 |
|
|
00A5 |
311 |
|
|
2116 |
312 |
|
|
EOH |
313 |
|
|
} |
314 |
|
|
|
315 |
|
|
sub InJISX4051SeparationSymbol { |
316 |
|
|
<<EOH; |
317 |
|
|
0021 |
318 |
|
|
003F |
319 |
|
|
EOH |
320 |
|
|
} |
321 |
|
|
|
322 |
|
|
sub InJISX4051UnseparatableCharacter { |
323 |
|
|
<<EOH; |
324 |
|
|
2014 |
325 |
|
|
2024 2026 |
326 |
|
|
EOH |
327 |
|
|
} |
328 |
|
|
|
329 |
|
|
sub InJISX4051WarichuuCloseBracket { |
330 |
|
|
<<EOH; |
331 |
|
|
0029 |
332 |
|
|
005D |
333 |
|
|
3015 |
334 |
|
|
EOH |
335 |
|
|
} |
336 |
|
|
|
337 |
|
|
sub InJISX4051WarichuuOpenBracket { |
338 |
|
|
<<EOH; |
339 |
|
|
0028 |
340 |
|
|
005B |
341 |
|
|
3014 |
342 |
|
|
EOH |
343 |
|
|
} |
344 |
|
|
|
345 |
|
|
=head1 COLLECTION NAMES |
346 |
|
|
|
347 |
|
|
=over 4 |
348 |
|
|
|
349 |
|
|
=item C<InJISX4051AlphabeticExceptSpace> |
350 |
|
|
|
351 |
|
|
(18) 欧文間隔以外の欧文用文字 |
352 |
|
|
|
353 |
|
|
=item C<InJISX4051AlphabeticSpace> |
354 |
|
|
|
355 |
|
|
(17) 欧文間隔 |
356 |
|
|
|
357 |
|
|
=item C<InJISX4051CharacterInDigits> |
358 |
|
|
|
359 |
|
|
(15) 連数字中の文字 |
360 |
|
|
|
361 |
|
|
=item C<InJISX4051CharacterInUnitSymbol> |
362 |
|
|
|
363 |
|
|
(16) 単位記号中の文字 |
364 |
|
|
|
365 |
|
|
=item C<InJISX4051CloseBracket> |
366 |
|
|
|
367 |
|
|
(2) 終わり括弧類 |
368 |
|
|
|
369 |
|
|
=item C<InJISX4051FullStop> |
370 |
|
|
|
371 |
|
|
(6) 句点類 |
372 |
|
|
|
373 |
|
|
=item C<InJISX4051Hiragana> |
374 |
|
|
|
375 |
|
|
(11) 平仮名 |
376 |
|
|
|
377 |
|
|
=item C<InJISX4051JapaneseSpace> |
378 |
|
|
|
379 |
|
|
(10) 和字間隔 |
380 |
|
|
|
381 |
|
|
=item C<InJISX4051LineStartKinsokuJapaneseCharacter> |
382 |
|
|
|
383 |
|
|
(3) 行頭禁則和字 |
384 |
|
|
|
385 |
|
|
=item C<InJISX4051MiddleDot> |
386 |
|
|
|
387 |
|
|
(5) 中点類 |
388 |
|
|
|
389 |
|
|
=item C<InJISX4051MiscJapaneseCharacter> |
390 |
|
|
|
391 |
|
|
(12) (1)〜(11)以外の和字 |
392 |
|
|
|
393 |
|
|
=item C<InJISX4051OpenBracket> |
394 |
|
|
|
395 |
|
|
(1) 始め括弧類 |
396 |
|
|
|
397 |
|
|
=item C<InJISX4051PostShortenedSymbol> |
398 |
|
|
|
399 |
|
|
(9) 後置省略記号 |
400 |
|
|
|
401 |
|
|
=item C<InJISX4051PreShortenedSymbol> |
402 |
|
|
|
403 |
|
|
(8) 前置省略記号 |
404 |
|
|
|
405 |
|
|
=item C<InJISX4051SeparationSymbol> |
406 |
|
|
|
407 |
|
|
(4) 区切り約物 |
408 |
|
|
|
409 |
|
|
=item C<InJISX4051UnseparatableCharacter> |
410 |
|
|
|
411 |
|
|
(7) 分離禁止文字 |
412 |
|
|
|
413 |
|
|
=item C<InJISX4051WarichuuCloseBracket> |
414 |
|
|
|
415 |
|
|
(20) 割注終わり括弧類 |
416 |
|
|
|
417 |
|
|
=item C<InJISX4051WarichuuOpenBracket> |
418 |
|
|
|
419 |
|
|
(19) 割注始め括弧類 |
420 |
|
|
|
421 |
|
|
=back |
422 |
|
|
|
423 |
|
|
=cut |
424 |
|
|
|
425 |
|
|
|
426 |
|
|
=head1 EXAMPLE |
427 |
|
|
|
428 |
|
|
use Char::Class::JISX4051; |
429 |
|
|
if ($s =~ /\p{InCharacterInUnitSymbol}/) { |
430 |
|
|
print "Match!\n"; |
431 |
|
|
} |
432 |
|
|
|
433 |
|
|
=head1 SEE ALSO |
434 |
|
|
|
435 |
|
|
JIS X 4051-1995, 『日本語文書の行組版方法 Line composition rules |
436 |
|
|
for Japanese documents』, 日本工業標準調査会 Japanese Industrial |
437 |
|
|
Standard Commitee (JISC <http://www.jisc.go.jp/>), 1995. |
438 |
|
|
|
439 |
|
|
=head1 LICENSE |
440 |
|
|
|
441 |
|
|
Copyright 2007 わかば Wakaba <w@suika.fam.cx> |
442 |
|
|
|
443 |
|
|
This library and the library generated by it is free software; |
444 |
|
|
you can redistribute them and/or modify them under the same |
445 |
|
|
terms as Perl itself. |
446 |
|
|
|
447 |
|
|
=cut |
448 |
|
|
|
449 |
|
|
1; |
450 |
|
|
### JISX4051.pm ends here |