ext/Encode/lib/Encode/JP/H2Z.pm

   1 #
   2 # $Id: H2Z.pm,v 2.2 2006/06/03 20:28:48 dankogai Exp dankogai $
   3 #
   4
   5 package Encode::JP::H2Z;
   6
   7 use strict;
   8 use warnings;
   9
  10 our $RCSID = q$Id: H2Z.pm,v 2.2 2006/06/03 20:28:48 dankogai Exp dankogai $;
  11 our $VERSION = do { my @r = ( q$Revision: 2.2 $ =~ /\d+/g ); sprintf "%d." . "%02d" x $#r, @r };
  12
  13 use Encode::CJKConstants qw(:all);
  14
  15 use vars qw(%_D2Z  $_PAT_D2Z
  16   %_Z2D  $_PAT_Z2D
  17   %_H2Z  $_PAT_H2Z
  18   %_Z2H  $_PAT_Z2H);
  19
  20 %_H2Z = (
  21     "\x8e\xa1" => "\xa1\xa3",    #¡£
  22     "\x8e\xa2" => "\xa1\xd6",    #¡Ö
  23     "\x8e\xa3" => "\xa1\xd7",    #¡×
  24     "\x8e\xa4" => "\xa1\xa2",    #¡¢
  25     "\x8e\xa5" => "\xa1\xa6",    #¡¦
  26     "\x8e\xa6" => "\xa5\xf2",    #¥ò
  27     "\x8e\xa7" => "\xa5\xa1",    #¥¡
  28     "\x8e\xa8" => "\xa5\xa3",    #¥£
  29     "\x8e\xa9" => "\xa5\xa5",    #¥¥
  30     "\x8e\xaa" => "\xa5\xa7",    #¥§
  31     "\x8e\xab" => "\xa5\xa9",    #¥©
  32     "\x8e\xac" => "\xa5\xe3",    #¥ã
  33     "\x8e\xad" => "\xa5\xe5",    #¥å
  34     "\x8e\xae" => "\xa5\xe7",    #¥ç
  35     "\x8e\xaf" => "\xa5\xc3",    #¥Ã
  36     "\x8e\xb0" => "\xa1\xbc",    #¡¼
  37     "\x8e\xb1" => "\xa5\xa2",    #¥¢
  38     "\x8e\xb2" => "\xa5\xa4",    #¥¤
  39     "\x8e\xb3" => "\xa5\xa6",    #¥¦
  40     "\x8e\xb4" => "\xa5\xa8",    #¥¨
  41     "\x8e\xb5" => "\xa5\xaa",    #¥ª
  42     "\x8e\xb6" => "\xa5\xab",    #¥«
  43     "\x8e\xb7" => "\xa5\xad",    #¥
  44     "\x8e\xb8" => "\xa5\xaf",    #¥¯
  45     "\x8e\xb9" => "\xa5\xb1",    #¥±
  46     "\x8e\xba" => "\xa5\xb3",    #¥³
  47     "\x8e\xbb" => "\xa5\xb5",    #¥µ
  48     "\x8e\xbc" => "\xa5\xb7",    #¥·
  49     "\x8e\xbd" => "\xa5\xb9",    #¥¹
  50     "\x8e\xbe" => "\xa5\xbb",    #¥»
  51     "\x8e\xbf" => "\xa5\xbd",    #¥½
  52     "\x8e\xc0" => "\xa5\xbf",    #¥¿
  53     "\x8e\xc1" => "\xa5\xc1",    #¥Á
  54     "\x8e\xc2" => "\xa5\xc4",    #¥Ä
  55     "\x8e\xc3" => "\xa5\xc6",    #¥Æ
  56     "\x8e\xc4" => "\xa5\xc8",    #¥È
  57     "\x8e\xc5" => "\xa5\xca",    #¥Ê
  58     "\x8e\xc6" => "\xa5\xcb",    #¥Ë
  59     "\x8e\xc7" => "\xa5\xcc",    #¥Ì
  60     "\x8e\xc8" => "\xa5\xcd",    #¥Í
  61     "\x8e\xc9" => "\xa5\xce",    #¥Î
  62     "\x8e\xca" => "\xa5\xcf",    #¥Ï
  63     "\x8e\xcb" => "\xa5\xd2",    #¥Ò
  64     "\x8e\xcc" => "\xa5\xd5",    #¥Õ
  65     "\x8e\xcd" => "\xa5\xd8",    #¥Ø
  66     "\x8e\xce" => "\xa5\xdb",    #¥Û
  67     "\x8e\xcf" => "\xa5\xde",    #¥Þ
  68     "\x8e\xd0" => "\xa5\xdf",    #¥ß
  69     "\x8e\xd1" => "\xa5\xe0",    #¥à
  70     "\x8e\xd2" => "\xa5\xe1",    #¥á
  71     "\x8e\xd3" => "\xa5\xe2",    #¥â
  72     "\x8e\xd4" => "\xa5\xe4",    #¥ä
  73     "\x8e\xd5" => "\xa5\xe6",    #¥æ
  74     "\x8e\xd6" => "\xa5\xe8",    #¥è
  75     "\x8e\xd7" => "\xa5\xe9",    #¥é
  76     "\x8e\xd8" => "\xa5\xea",    #¥ê
  77     "\x8e\xd9" => "\xa5\xeb",    #¥ë
  78     "\x8e\xda" => "\xa5\xec",    #¥ì
  79     "\x8e\xdb" => "\xa5\xed",    #¥í
  80     "\x8e\xdc" => "\xa5\xef",    #¥ï
  81     "\x8e\xdd" => "\xa5\xf3",    #¥ó
  82     "\x8e\xde" => "\xa1\xab",    #¡«
  83     "\x8e\xdf" => "\xa1\xac",    #¡¬
  84 );
  85
  86 %_D2Z = (
  87     "\x8e\xb6\x8e\xde" => "\xa5\xac",    #¥¬
  88     "\x8e\xb7\x8e\xde" => "\xa5\xae",    #¥®
  89     "\x8e\xb8\x8e\xde" => "\xa5\xb0",    #¥°
  90     "\x8e\xb9\x8e\xde" => "\xa5\xb2",    #¥²
  91     "\x8e\xba\x8e\xde" => "\xa5\xb4",    #¥´
  92     "\x8e\xbb\x8e\xde" => "\xa5\xb6",    #¥¶
  93     "\x8e\xbc\x8e\xde" => "\xa5\xb8",    #¥¸
  94     "\x8e\xbd\x8e\xde" => "\xa5\xba",    #¥º
  95     "\x8e\xbe\x8e\xde" => "\xa5\xbc",    #¥¼
  96     "\x8e\xbf\x8e\xde" => "\xa5\xbe",    #¥¾
  97     "\x8e\xc0\x8e\xde" => "\xa5\xc0",    #¥À
  98     "\x8e\xc1\x8e\xde" => "\xa5\xc2",    #¥Â
  99     "\x8e\xc2\x8e\xde" => "\xa5\xc5",    #¥Å
 100     "\x8e\xc3\x8e\xde" => "\xa5\xc7",    #¥Ç
 101     "\x8e\xc4\x8e\xde" => "\xa5\xc9",    #¥É
 102     "\x8e\xca\x8e\xde" => "\xa5\xd0",    #¥Ð
 103     "\x8e\xcb\x8e\xde" => "\xa5\xd3",    #¥Ó
 104     "\x8e\xcc\x8e\xde" => "\xa5\xd6",    #¥Ö
 105     "\x8e\xcd\x8e\xde" => "\xa5\xd9",    #¥Ù
 106     "\x8e\xce\x8e\xde" => "\xa5\xdc",    #¥Ü
 107     "\x8e\xca\x8e\xdf" => "\xa5\xd1",    #¥Ñ
 108     "\x8e\xcb\x8e\xdf" => "\xa5\xd4",    #¥Ô
 109     "\x8e\xcc\x8e\xdf" => "\xa5\xd7",    #¥×
 110     "\x8e\xcd\x8e\xdf" => "\xa5\xda",    #¥Ú
 111     "\x8e\xce\x8e\xdf" => "\xa5\xdd",    #¥Ý
 112     "\x8e\xb3\x8e\xde" => "\xa5\xf4",    #¥ô
 113 );
 114
 115 # init only once;
 116
 117 #$_PAT_D2Z = join("|", keys %_D2Z);
 118 #$_PAT_H2Z = join("|", keys %_H2Z);
 119
 120 %_Z2H = reverse %_H2Z;
 121 %_Z2D = reverse %_D2Z;
 122
 123 #$_PAT_Z2H    = join("|", keys %_Z2H);
 124 #$_PAT_Z2D    = join("|", keys %_Z2D);
 125
 126 sub h2z {
 127     no warnings qw(uninitialized);
 128     my $r_str          = shift;
 129     my ($keep_dakuten) = @_;
 130     my $n              = 0;
 131     unless ($keep_dakuten) {
 132         $n = (
 133             $$r_str =~ s(
 134                ($RE{EUC_KANA}
 135                 (?:\x8e[\xde\xdf])?)
 136                ){
 137           my $str = $1;
 138           $_D2Z{$str} || $_H2Z{$str} ||
 139               # in case dakuten and handakuten are side-by-side!
 140               $_H2Z{substr($str,0,2)} . $_H2Z{substr($str,2,2)};
 141           }eogx
 142         );
 143     }
 144     else {
 145         $n = (
 146             $$r_str =~ s(
 147                ($RE{EUC_KANA})
 148                ){
 149           $_H2Z{$1};
 150           }eogx
 151         );
 152     }
 153     $n;
 154 }
 155
 156 sub z2h {
 157     my $r_str = shift;
 158     my $n     = (
 159         $$r_str =~ s(
 160               ($RE{EUC_C}|$RE{EUC_0212}|$RE{EUC_KANA})
 161               ){
 162          $_Z2D{$1} || $_Z2H{$1} || $1;
 163          }eogx
 164     );
 165     $n;
 166 }
 167
 168 1;
 169 __END__
 170
 171
 172 =head1 NAME
 173
 174 Encode::JP::H2Z -- internally used by Encode::JP::2022_JP*
 175
 176 =cut