Commit | Line | Data |
0e567a6c |
1 | # |
64ffdd5e |
2 | # $Id: H2Z.pm,v 0.99 2002/03/25 19:29:22 dankogai Exp dankogai $ |
0e567a6c |
3 | # |
4 | |
5 | package Encode::JP::H2Z; |
6 | |
7 | use strict; |
0e567a6c |
8 | |
64ffdd5e |
9 | our $RCSID = q$Id: H2Z.pm,v 0.99 2002/03/25 19:29:22 dankogai Exp dankogai $; |
10 | our $VERSION = do { my @r = (q$Revision: 0.99 $ =~ /\d+/g); sprintf "%d."."%02d" x $#r, @r }; |
0e567a6c |
11 | |
12 | use Carp; |
13 | |
64ffdd5e |
14 | use Encode::CJKConstants qw(:all); |
0e567a6c |
15 | |
16 | use vars qw(%_D2Z $_PAT_D2Z |
17 | %_Z2D $_PAT_Z2D |
18 | %_H2Z $_PAT_H2Z |
19 | %_Z2H $_PAT_Z2H); |
20 | |
21 | %_H2Z = ( |
22 | "\x8e\xa1" => "\xa1\xa3", #¡£ |
23 | "\x8e\xa2" => "\xa1\xd6", #¡Ö |
24 | "\x8e\xa3" => "\xa1\xd7", #¡× |
25 | "\x8e\xa4" => "\xa1\xa2", #¡¢ |
26 | "\x8e\xa5" => "\xa1\xa6", #¡¦ |
27 | "\x8e\xa6" => "\xa5\xf2", #¥ò |
28 | "\x8e\xa7" => "\xa5\xa1", #¥¡ |
29 | "\x8e\xa8" => "\xa5\xa3", #¥£ |
30 | "\x8e\xa9" => "\xa5\xa5", #¥¥ |
31 | "\x8e\xaa" => "\xa5\xa7", #¥§ |
32 | "\x8e\xab" => "\xa5\xa9", #¥© |
33 | "\x8e\xac" => "\xa5\xe3", #¥ã |
34 | "\x8e\xad" => "\xa5\xe5", #¥å |
35 | "\x8e\xae" => "\xa5\xe7", #¥ç |
36 | "\x8e\xaf" => "\xa5\xc3", #¥Ã |
37 | "\x8e\xb0" => "\xa1\xbc", #¡¼ |
38 | "\x8e\xb1" => "\xa5\xa2", #¥¢ |
39 | "\x8e\xb2" => "\xa5\xa4", #¥¤ |
40 | "\x8e\xb3" => "\xa5\xa6", #¥¦ |
41 | "\x8e\xb4" => "\xa5\xa8", #¥¨ |
42 | "\x8e\xb5" => "\xa5\xaa", #¥ª |
43 | "\x8e\xb6" => "\xa5\xab", #¥« |
44 | "\x8e\xb7" => "\xa5\xad", #¥ |
45 | "\x8e\xb8" => "\xa5\xaf", #¥¯ |
46 | "\x8e\xb9" => "\xa5\xb1", #¥± |
47 | "\x8e\xba" => "\xa5\xb3", #¥³ |
48 | "\x8e\xbb" => "\xa5\xb5", #¥µ |
49 | "\x8e\xbc" => "\xa5\xb7", #¥· |
50 | "\x8e\xbd" => "\xa5\xb9", #¥¹ |
51 | "\x8e\xbe" => "\xa5\xbb", #¥» |
52 | "\x8e\xbf" => "\xa5\xbd", #¥½ |
53 | "\x8e\xc0" => "\xa5\xbf", #¥¿ |
54 | "\x8e\xc1" => "\xa5\xc1", #¥Á |
55 | "\x8e\xc2" => "\xa5\xc4", #¥Ä |
56 | "\x8e\xc3" => "\xa5\xc6", #¥Æ |
57 | "\x8e\xc4" => "\xa5\xc8", #¥È |
58 | "\x8e\xc5" => "\xa5\xca", #¥Ê |
59 | "\x8e\xc6" => "\xa5\xcb", #¥Ë |
60 | "\x8e\xc7" => "\xa5\xcc", #¥Ì |
61 | "\x8e\xc8" => "\xa5\xcd", #¥Í |
62 | "\x8e\xc9" => "\xa5\xce", #¥Î |
63 | "\x8e\xca" => "\xa5\xcf", #¥Ï |
64 | "\x8e\xcb" => "\xa5\xd2", #¥Ò |
65 | "\x8e\xcc" => "\xa5\xd5", #¥Õ |
66 | "\x8e\xcd" => "\xa5\xd8", #¥Ø |
67 | "\x8e\xce" => "\xa5\xdb", #¥Û |
68 | "\x8e\xcf" => "\xa5\xde", #¥Þ |
69 | "\x8e\xd0" => "\xa5\xdf", #¥ß |
70 | "\x8e\xd1" => "\xa5\xe0", #¥à |
71 | "\x8e\xd2" => "\xa5\xe1", #¥á |
72 | "\x8e\xd3" => "\xa5\xe2", #¥â |
73 | "\x8e\xd4" => "\xa5\xe4", #¥ä |
74 | "\x8e\xd5" => "\xa5\xe6", #¥æ |
75 | "\x8e\xd6" => "\xa5\xe8", #¥è |
76 | "\x8e\xd7" => "\xa5\xe9", #¥é |
77 | "\x8e\xd8" => "\xa5\xea", #¥ê |
78 | "\x8e\xd9" => "\xa5\xeb", #¥ë |
79 | "\x8e\xda" => "\xa5\xec", #¥ì |
80 | "\x8e\xdb" => "\xa5\xed", #¥í |
81 | "\x8e\xdc" => "\xa5\xef", #¥ï |
82 | "\x8e\xdd" => "\xa5\xf3", #¥ó |
83 | "\x8e\xde" => "\xa1\xab", #¡« |
84 | "\x8e\xdf" => "\xa1\xac", #¡¬ |
85 | ); |
86 | |
87 | %_D2Z = ( |
88 | "\x8e\xb6\x8e\xde" => "\xa5\xac", #¥¬ |
89 | "\x8e\xb7\x8e\xde" => "\xa5\xae", #¥® |
90 | "\x8e\xb8\x8e\xde" => "\xa5\xb0", #¥° |
91 | "\x8e\xb9\x8e\xde" => "\xa5\xb2", #¥² |
92 | "\x8e\xba\x8e\xde" => "\xa5\xb4", #¥´ |
93 | "\x8e\xbb\x8e\xde" => "\xa5\xb6", #¥¶ |
94 | "\x8e\xbc\x8e\xde" => "\xa5\xb8", #¥¸ |
95 | "\x8e\xbd\x8e\xde" => "\xa5\xba", #¥º |
96 | "\x8e\xbe\x8e\xde" => "\xa5\xbc", #¥¼ |
97 | "\x8e\xbf\x8e\xde" => "\xa5\xbe", #¥¾ |
98 | "\x8e\xc0\x8e\xde" => "\xa5\xc0", #¥À |
99 | "\x8e\xc1\x8e\xde" => "\xa5\xc2", #¥Â |
100 | "\x8e\xc2\x8e\xde" => "\xa5\xc5", #¥Å |
101 | "\x8e\xc3\x8e\xde" => "\xa5\xc7", #¥Ç |
102 | "\x8e\xc4\x8e\xde" => "\xa5\xc9", #¥É |
103 | "\x8e\xca\x8e\xde" => "\xa5\xd0", #¥Ð |
104 | "\x8e\xcb\x8e\xde" => "\xa5\xd3", #¥Ó |
105 | "\x8e\xcc\x8e\xde" => "\xa5\xd6", #¥Ö |
106 | "\x8e\xcd\x8e\xde" => "\xa5\xd9", #¥Ù |
107 | "\x8e\xce\x8e\xde" => "\xa5\xdc", #¥Ü |
108 | "\x8e\xca\x8e\xdf" => "\xa5\xd1", #¥Ñ |
109 | "\x8e\xcb\x8e\xdf" => "\xa5\xd4", #¥Ô |
110 | "\x8e\xcc\x8e\xdf" => "\xa5\xd7", #¥× |
111 | "\x8e\xcd\x8e\xdf" => "\xa5\xda", #¥Ú |
112 | "\x8e\xce\x8e\xdf" => "\xa5\xdd", #¥Ý |
113 | "\x8e\xb3\x8e\xde" => "\xa5\xf4", #¥ô |
114 | ); |
115 | |
116 | # init only once; |
117 | |
118 | #$_PAT_D2Z = join("|", keys %_D2Z); |
119 | #$_PAT_H2Z = join("|", keys %_H2Z); |
120 | |
121 | %_Z2H = reverse %_H2Z; |
122 | %_Z2D = reverse %_D2Z; |
123 | |
124 | #$_PAT_Z2H = join("|", keys %_Z2H); |
125 | #$_PAT_Z2D = join("|", keys %_Z2D); |
126 | |
127 | sub h2z { |
128 | my $r_str = shift; |
129 | my ($keep_dakuten) = @_; |
130 | my $n = 0; |
131 | unless ($keep_dakuten){ |
132 | $n = ( |
133 | $$r_str =~ s( |
134 | ($RE{EUC_KANA} |
135 | (?:\x8e[\xde\xdf])?) |
136 | ){ |
137 | my $str = $1; |
138 | $_D2Z{$str} || $_H2Z{$str} || |
139 | # in case dakuten and handakuten are side-by-side! |
140 | $_H2Z{substr($str,0,2)} . $_H2Z{substr($str,2,2)}; |
141 | }eogx |
142 | ); |
143 | }else{ |
144 | $n = ( |
145 | $$r_str =~ s( |
146 | ($RE{EUC_KANA}) |
147 | ){ |
148 | $_H2Z{$1}; |
149 | }eogx |
150 | ); |
151 | } |
152 | $n; |
153 | } |
154 | |
155 | sub z2h { |
156 | my $r_str = shift; |
157 | my $n = ( |
158 | $$r_str =~ s( |
159 | ($RE{EUC_C}|$RE{EUC_0212}|$RE{EUC_KANA}) |
160 | ){ |
161 | $_Z2D{$1} || $_Z2H{$1} || $1; |
162 | }eogx |
163 | ); |
164 | $n; |
165 | } |
166 | |
167 | 1; |