1 |
wakaba |
1.1 |
#!/usr/local/bin/perl
|
2 |
|
|
use utf8; ## This file is written in UTF-8
|
3 |
|
|
use strict;
|
4 |
|
|
require 'mkpm.pl';
|
5 |
|
|
use vars qw(%PROP %SET %SET_ALIAS);
|
6 |
|
|
$PROP{module_name} = 'JISX4051';
|
7 |
|
|
$PROP{version} = do{my @r=(q$Revision: 1.1 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};
|
8 |
|
|
$PROP{author_name} = 'わかば Wakaba';
|
9 |
|
|
$PROP{author_mail} = 'w@suika.fam.cx';
|
10 |
|
|
|
11 |
|
|
$PROP{pod_description} = <<EOH;
|
12 |
|
|
JIS X 4051 character classes represented in UCS,
|
13 |
|
|
defined by JIS X 4051 Appendix
|
14 |
|
|
EOH
|
15 |
|
|
|
16 |
|
|
$PROP{pod_see_also} = <<EOH;
|
17 |
|
|
JIS X 4051-1995, 『日本語文書の行組版方法 Line composition rules
|
18 |
|
|
for Japanese documents』, 日本工業標準調査会 Japanese Industrial
|
19 |
|
|
Standard Commitee (JISC <http://www.jisc.go.jp/>), 1995.
|
20 |
|
|
EOH
|
21 |
|
|
|
22 |
|
|
$SET{OpenBracket} = col2list (<<EOH);
|
23 |
|
|
#DESCRIPTION (1) 始め括弧類
|
24 |
|
|
00 28 5B 7B
|
25 |
|
|
20 18 1B 1C 1F
|
26 |
|
|
30 08 0A 0C 0E 10 14 16 18 1A 1D
|
27 |
|
|
EOH
|
28 |
|
|
|
29 |
|
|
$SET{CloseBracket} = col2list (<<EOH);
|
30 |
|
|
#DESCRIPTION (2) 終わり括弧類
|
31 |
|
|
00 29 2C 5D 7D
|
32 |
|
|
20 19 1A 1D 1E
|
33 |
|
|
30 01 09 0B 0D 0F 11 15 17 19 1B 1E 1F
|
34 |
|
|
EOH
|
35 |
|
|
|
36 |
|
|
$SET{LineStartKinsokuJapaneseCharacter} = col2list (<<EOH);
|
37 |
|
|
#DESCRIPTION (3) 行頭禁則和字
|
38 |
|
|
20 3C 44
|
39 |
|
|
30 1C 41 43 45 47 49 63 83 85 87 8E 9D 9E
|
40 |
|
|
30 A1 A3 A5 A7 A9 C3 E3 E5 E7 EE F5 F6 FC FD FE
|
41 |
|
|
EOH
|
42 |
|
|
## U+2044: 処理系定義 implemention-defined
|
43 |
|
|
## U+30FC: 処理系定義
|
44 |
|
|
## U+3041-U+308E: 処理系定義。このクラスから削除した場合は
|
45 |
|
|
## 平仮名クラスに編入。 When these characters
|
46 |
|
|
## are removed from this class, they shall be included
|
47 |
|
|
## in Hiragana class. (JIS X 4051-1995 4.1.1)
|
48 |
|
|
|
49 |
|
|
$SET{SeparationSymbol} = col2list (<<EOH);
|
50 |
|
|
#DESCRIPTION (4) 区切り約物
|
51 |
|
|
00 21 3F
|
52 |
|
|
EOH
|
53 |
|
|
|
54 |
|
|
$SET{MiddleDot} = col2list (<<EOH);
|
55 |
|
|
#DESCRIPTION (5) 中点類
|
56 |
|
|
00 3A 3B FB
|
57 |
|
|
EOH
|
58 |
|
|
|
59 |
|
|
$SET{FullStop} = col2list (<<EOH);
|
60 |
|
|
#DESCRIPTION (6) 句点類
|
61 |
|
|
00 2E
|
62 |
|
|
30 02
|
63 |
|
|
EOH
|
64 |
|
|
|
65 |
|
|
$SET{UnseparatableCharacter} = col2list (<<EOH);
|
66 |
|
|
#DESCRIPTION (7) 分離禁止文字
|
67 |
|
|
20 14 24-26
|
68 |
|
|
EOH
|
69 |
|
|
|
70 |
|
|
$SET{PreShortenedSymbol} = col2list (<<EOH);
|
71 |
|
|
#DESCRIPTION (8) 前置省略記号
|
72 |
|
|
00 24 A3 A5
|
73 |
|
|
21 16
|
74 |
|
|
EOH
|
75 |
|
|
|
76 |
|
|
$SET{PostShortenedSymbol} = col2list (<<EOH);
|
77 |
|
|
#DESCRIPTION (9) 後置省略記号
|
78 |
|
|
00 25 A2 B0
|
79 |
|
|
20 30-33
|
80 |
|
|
EOH
|
81 |
|
|
|
82 |
|
|
$SET{JapaneseSpace} = <<EOH;
|
83 |
|
|
#DESCRIPTION (10) 和字間隔
|
84 |
|
|
\x{3000}
|
85 |
|
|
EOH
|
86 |
|
|
|
87 |
|
|
$SET{Hiragana} = col2list (<<EOH);
|
88 |
|
|
#DESCRIPTION (11) 平仮名
|
89 |
|
|
30 42 44 46 48 4A-62 64-82 84 86 88-8D 8F-94
|
90 |
|
|
EOH
|
91 |
|
|
|
92 |
|
|
$SET{MiscJapaneseCharacter} = col2list (<<EOH).<<EOH;
|
93 |
|
|
#DESCRIPTION (12) (1)〜(11)以外の和字
|
94 |
|
|
00 2B 2D 3C 3D 3E A7 A9 AE B1 B6 D7 F7
|
95 |
|
|
20 3B
|
96 |
|
|
21 60-7F 90-EA
|
97 |
|
|
24 60-EA
|
98 |
|
|
25 00-7F 80-95 A0-EF
|
99 |
|
|
26 00-13 1A-6E
|
100 |
|
|
27 01-04 06-09 0C-27 29-4B 4D 4F-52 56 58-5E 61-67 76-94 98-AF B1-BE
|
101 |
|
|
30 03 04 06 07 12 13 20 36 A2 A4 A6 A8 AA-C2 C4-E2 E4 E6 E8-ED EF-F4 F7-FA
|
102 |
|
|
32 20-43 80-B0 D0-FE
|
103 |
|
|
33 00-57 71-76 80-DD
|
104 |
|
|
EOH
|
105 |
|
|
!4E00 9FA5
|
106 |
|
|
EOH
|
107 |
|
|
|
108 |
|
|
## (13) 添え字付き親文字群中の文字
|
109 |
|
|
## (14) ルビ付き親文字群中の文字
|
110 |
|
|
## -- 文字指定無し。
|
111 |
|
|
|
112 |
|
|
$SET{CharacterInDigits} = <<EOH;
|
113 |
|
|
#DESCRIPTION (15) 連数字中の文字
|
114 |
|
|
!0030 0039
|
115 |
|
|
\x{0020}\x{002C}\x{002E}
|
116 |
|
|
EOH
|
117 |
|
|
|
118 |
|
|
$SET{CharacterInUnitSymbol} = <<EOH;
|
119 |
|
|
#DESCRIPTION (16) 単位記号中の文字
|
120 |
|
|
!0041 005A
|
121 |
|
|
!0061 007A
|
122 |
|
|
!2126
|
123 |
|
|
EOH
|
124 |
|
|
|
125 |
|
|
$SET{AlphabeticSpace} = <<EOH;
|
126 |
|
|
#DESCRIPTION (17) 欧文間隔
|
127 |
|
|
!0020
|
128 |
|
|
EOH
|
129 |
|
|
|
130 |
|
|
$SET{AlphabeticExceptSpace} = col2list (<<EOH);
|
131 |
|
|
#DESCRIPTION (18) 欧文間隔以外の欧文用文字
|
132 |
|
|
00 21-7E
|
133 |
|
|
00 A1-FF
|
134 |
|
|
01 00-7F
|
135 |
|
|
02 50-A8
|
136 |
|
|
03 74-75 7A 7E 84-8A 8C 8E-A1 A3-CE
|
137 |
|
|
04 01-0C 0E-4F 51-5C 5E-7F 80-86 90-C4 C7-C8 CB-CC D0-EB EE-F5 F8-F9
|
138 |
|
|
20 00-2E 30-46 70 74-8E A0-AA
|
139 |
|
|
21 00-38 53-82 90-EA
|
140 |
|
|
22 00-7F 80-F1
|
141 |
|
|
23 12
|
142 |
|
|
24 60-EA
|
143 |
|
|
25 00-7F 80-95 A0-EF
|
144 |
|
|
26 00-13 1A-6F
|
145 |
|
|
27 01-04 06-09 0C-27 29-4B 4D 4F-52 56 58-5E 61-67 76-94 98-AF B1-BE
|
146 |
|
|
EOH
|
147 |
|
|
|
148 |
|
|
$SET{WarichuuOpenBracket} = col2list (<<EOH);
|
149 |
|
|
#DESCRIPTION (19) 割注始め括弧類
|
150 |
|
|
00 28 5B
|
151 |
|
|
30 14
|
152 |
|
|
EOH
|
153 |
|
|
|
154 |
|
|
$SET{WarichuuCloseBracket} = col2list (<<EOH);
|
155 |
|
|
#DESCRIPTION (20) 割注終わり括弧類
|
156 |
|
|
00 29 5D
|
157 |
|
|
30 15
|
158 |
|
|
EOH
|
159 |
|
|
|
160 |
|
|
&print_module;
|
161 |
|
|
|
162 |
|
|
## $Date: 2002/08/31 11:17:10 $
|
163 |
|
|
### JISX4051-src.upl ends here
|