| 1 |
#!/usr/bin/perl
|
| 2 |
|
| 3 |
=head1 NAME
|
| 4 |
|
| 5 |
typo2tbl --- Mapping table converter, Microsoft typography's table
|
| 6 |
to our tbl format
|
| 7 |
|
| 8 |
=head1 USAGE
|
| 9 |
|
| 10 |
$ perl typo2tbl.pl 932.txt > ms932.tbl
|
| 11 |
|
| 12 |
=cut
|
| 13 |
|
| 14 |
use v5.8.0;
|
| 15 |
use strict;
|
| 16 |
my %add_hdr = (
|
| 17 |
932 => q(# ucm:code_set_alias Windows-31J Windows31J csWindows31J SJIS-CP932 MS932 X-MS-CP932 mscode X-SJIS-CP932 sjis-ms),
|
| 18 |
);
|
| 19 |
|
| 20 |
my %U2C;
|
| 21 |
my %C2U;
|
| 22 |
my @hdr;
|
| 23 |
my $u = 0;
|
| 24 |
my $Umode = 0;
|
| 25 |
while (<>) {
|
| 26 |
if ($Umode) {
|
| 27 |
if (/0x(\S+) 0x(\S+)/) {
|
| 28 |
$U2C{ hex $1 } = hex $2;
|
| 29 |
}
|
| 30 |
} elsif (/0x\S+ 0x\S+ ;Lead Byte Range/) {
|
| 31 |
} elsif (/0x(\S+) 0x(\S+)/) {
|
| 32 |
$C2U{ $u + hex $1 } = hex $2;
|
| 33 |
} elsif (/DBCSTABLE.+?;LeadByte = 0x(\S+)/) {
|
| 34 |
$u = 0x100 * hex $1;
|
| 35 |
} elsif (/WCTABLE/) {
|
| 36 |
$Umode = 1;
|
| 37 |
} elsif (/^CODEPAGE (\d+)/) {
|
| 38 |
push @hdr, qq(# name ms$1);
|
| 39 |
push @hdr, $add_hdr{$1};
|
| 40 |
} elsif (/^CPINFO (\d) (0x\S+) (0x\S+)/) {
|
| 41 |
push @hdr, qq(# ucm:mb_cur_max $1);
|
| 42 |
push @hdr, qq(# <-ucs-substition $2);
|
| 43 |
push @hdr, qq(# ->ucs-substition $3);
|
| 44 |
}
|
| 45 |
}
|
| 46 |
|
| 47 |
{
|
| 48 |
my @name = split /\n/, require 'unicore/Name.pl';
|
| 49 |
my %name;
|
| 50 |
for (@name) {
|
| 51 |
if (/^(....) ([^\t]+)/) {
|
| 52 |
$name{hex $1} = $2;
|
| 53 |
}
|
| 54 |
}
|
| 55 |
sub NAME ($) {
|
| 56 |
$_[0] < 0x0020 ? '<control>' :
|
| 57 |
$_[0] < 0x007F ? $name{$_[0]} :
|
| 58 |
$_[0] < 0x00A0 ? '<control>' :
|
| 59 |
$name{$_[0]} ? $name{$_[0]} :
|
| 60 |
$_[0] < 0x00A0 ? '<control>' :
|
| 61 |
$_[0] < 0x3400 ? '' :
|
| 62 |
$_[0] < 0xA000 ? '<cjk>' :
|
| 63 |
$_[0] < 0xE000 ? '<hangul>' :
|
| 64 |
$_[0] < 0xF900 ? '<private>' :
|
| 65 |
'';
|
| 66 |
}
|
| 67 |
}
|
| 68 |
|
| 69 |
my @t;
|
| 70 |
for my $U (keys %U2C) {
|
| 71 |
if ($C2U{$U2C{$U}} == $U) {
|
| 72 |
push @t, sprintf '0x%04X U+%04X # %s',
|
| 73 |
$U2C{$U}, $U, NAME ($U);
|
| 74 |
delete $C2U{ $U2C{$U} };
|
| 75 |
} else {
|
| 76 |
push @t, sprintf '0x%04X U+%04X <- # %s',
|
| 77 |
$U2C{$U}, $U, NAME ($U);
|
| 78 |
}
|
| 79 |
}
|
| 80 |
for my $C (keys %C2U) {
|
| 81 |
push @t, sprintf '0x%04X U+%04X -> # %s',
|
| 82 |
$C, $C2U{$C}, NAME ($C2U{$C});
|
| 83 |
}
|
| 84 |
|
| 85 |
for (@t) {
|
| 86 |
s/^0x00(..)/0x$1/;
|
| 87 |
}
|
| 88 |
|
| 89 |
print "#; This file is auto-generated (at @{[ sprintf '%04d-%02d-%02dT%02d:%02d:%02dZ', (gmtime)[5]+1900, (gmtime)[4]+1, (gmtime)[3,2,1,0] ]}).\n#; Do not edit by hand!\n";
|
| 90 |
print join "\n", @hdr, sort @t;
|
| 91 |
print "\n";
|
| 92 |
|
| 93 |
=head1 SEE ALSO
|
| 94 |
|
| 95 |
Character sets and codepages (Microsoft typography)
|
| 96 |
<http://www.microsoft.com/typography/unicode/cscp.htm>
|
| 97 |
|
| 98 |
=head1 AUTHOR
|
| 99 |
|
| 100 |
Nanashi-san (SuikaWiki:WindowsCodePage
|
| 101 |
<http://suika.fam.cx/~wakaba/-temp/wiki/wiki?WindowsCodePage>)
|
| 102 |
|
| 103 |
=head1 LICENSE
|
| 104 |
|
| 105 |
Public Domain.
|
| 106 |
|
| 107 |
=cut
|
| 108 |
|
| 109 |
# $Date: 2002/11/02 12:10:45 $
|
Parent Directory
| 
Revision Log