Table/tool/tbl2pm.pl

#!/usr/bin/perl
use strict;
use 5.7.3;

=head1 NAME

tbl2pm --- Perl module generator from (completed) PETBL
table file

=head1 DESCRIPTION

This script generates dumped perl module file of convertion
table used with Encode::Table module.

=head1 USAGE

 $ ./tbl2pm.pl table.tbl > table.pm

=cut

require Encode;
my %T;
my %option;
while (<>) {
if (/^0x([0-9A-F]+)\tU\+([0-9A-F]+)\t/) {
  $T{ hex $1 } = hex $2;
} elsif (/^#\?o ([A-Za-z0-9-]+)="([^"]+)"/) {
  my $n = $1; $n =~ tr/-/_/; $option{$n} = $2;
}
}

require 'internal.pl';
$option{offset} ||= &internal::cp_start (%option) || $internal::cp_start{ $option{charset_name} || $option{table_name} };

my (@T, @U);
$option{_start} = $option{charset_chars} == 94 ? 0x21 : 0x20;
@U = sort { $a->[0] <=> $b->[0] }
     map { 
       [ ($option{charset_dimension} > 1 ?
          (int ($_ / 0x100) - $option{_start}) * $option{charset_chars} : 0)
         + (($_ % 0x100) - $option{_start}) + $option{offset},
         $T{ $_ } ]
     } keys %T;

my $tbl = '';
my $pack = '';
for (@U) {
  for ($_->[0], $_->[1]) {
    $_ = chr $_;
    Encode::_utf8_off ($_);
    $tbl .= $_;
    $pack .= 'a' . length ($_);
  }
}

print <<EOH;
## This file is auto-generated (at @{[ sprintf '%04d-%02d-%02dT%02d:%02d:%02dZ', (gmtime)[5]+1900, (gmtime)[4]+1, (gmtime)[3,2,1,0] ]}).
## Do not edit by hand!

=head1 NAME

Encode::Table::$option{table_name} --- Convertion tables
used with Encode::Table, C<$option{table_name}_to_ucs>
and C<ucs_to_$option{table_name}>

=head1 TABLES

=over 4

=item $option{table_name}_to_ucs

Convertion table of $option{table_name} -> ucs

=item ucs_to_$option{table_name}

Convertion table of ucs -> $option{table_name}

=back

=head1 SEE ALSO

Encode::Table

=head1 LICENSE

See source table of this module.  (It may be named as
C<$option{table_name}.tbr>.)

=cut

package Encode::Table::$option{table_name};
use strict;
our \$VERSION = q(@{[sprintf '%04d.%02d%02d', (gmtime)[5]+1900, (gmtime)[4]+1, (gmtime)[3]]});

## These tables are embeded in binary, so that your editor
## might break the data or might hang up.


#

our %L2U = map {Encode::_utf8_on (\$_) if length \$_ > 1; \$_} unpack
(q{$pack}, <<'END');
EOH

print $tbl,"\nEND\n";

print <<EOH;
our %U2L = reverse %L2U;
sub import {
  \$Encode::Table::TABLE{$option{table_name}_to_ucs} = \\%L2U;
  \$Encode::Table::TABLE{ucs_to_$option{table_name}} = \\%U2L;
}
1;
### $option{table_name}.pm ends here
EOH


=head1 SEE ALSO

Encode::Table, internal.pl

=head1 AUTHOR

Nanashi-san

=head1 LICENSE

Copyright 2002 AUTHOR

This library is free software; you can redistribute it
and/or modify it under the same terms as Perl itself.

AUTHOR does NOT claim any right to the data generated by
this script.  License of generated data fully depends
author of source data.

=cut

1; ## $Date: 2002/10/12 07:27:01 $
### tbl2pm.pl ends here
1	wakaba	1.1	#!/usr/bin/perl
2			use strict;
3			use 5.7.3;
4
5	wakaba	1.2	=head1 NAME
6
7			tbl2pm --- Perl module generator from (completed) PETBL
8			table file
9
10			=head1 DESCRIPTION
11
12			This script generates dumped perl module file of convertion
13			table used with Encode::Table module.
14
15			=head1 USAGE
16
17			$ ./tbl2pm.pl table.tbl > table.pm
18
19			=cut
20
21	wakaba	1.1	require Encode;
22			my %T;
23			my %option;
24			while (<>) {
25	wakaba	1.3	if (/^0x([0-9A-F]+)\tU\+([0-9A-F]+)\t/) {
26	wakaba	1.1	$T{ hex $1 } = hex $2;
27			} elsif (/^#\?o ([A-Za-z0-9-]+)="([^"]+)"/) {
28			my $n = $1; $n =~ tr/-/_/; $option{$n} = $2;
29			}
30			}
31
32			require 'internal.pl';
33			$option{offset} \|\|= &internal::cp_start (%option) \|\| $internal::cp_start{ $option{charset_name} \|\| $option{table_name} };
34
35			my (@T, @U);
36			$option{_start} = $option{charset_chars} == 94 ? 0x21 : 0x20;
37			@U = sort { $a->[0] <=> $b->[0] }
38			map {
39	wakaba	1.3	[ ($option{charset_dimension} > 1 ?
40			(int ($_ / 0x100) - $option{_start}) * $option{charset_chars} : 0)
41	wakaba	1.1	+ (($_ % 0x100) - $option{_start}) + $option{offset},
42			$T{ $_ } ]
43			} keys %T;
44
45			my $tbl = '';
46			my $pack = '';
47			for (@U) {
48			for ($_->[0], $_->[1]) {
49	wakaba	1.4	$_ = chr $_;
50	wakaba	1.1	Encode::_utf8_off ($_);
51			$tbl .= $_;
52			$pack .= 'a' . length ($_);
53			}
54			}
55
56			print <<EOH;
57			## This file is auto-generated (at @{[ sprintf '%04d-%02d-%02dT%02d:%02d:%02dZ', (gmtime)[5]+1900, (gmtime)[4]+1, (gmtime)[3,2,1,0] ]}).
58			## Do not edit by hand!
59
60			=head1 NAME
61
62			Encode::Table::$option{table_name} --- Convertion tables
63			used with Encode::Table, C<$option{table_name}_to_ucs>
64			and C<ucs_to_$option{table_name}>
65
66			=head1 TABLES
67
68			=over 4
69
70			=item $option{table_name}_to_ucs
71
72			Convertion table of $option{table_name} -> ucs
73
74			=item ucs_to_$option{table_name}
75
76			Convertion table of ucs -> $option{table_name}
77
78			=back
79
80			=head1 SEE ALSO
81
82			Encode::Table
83
84			=head1 LICENSE
85
86			See source table of this module. (It may be named as
87			C<$option{table_name}.tbr>.)
88
89			=cut
90
91			package Encode::Table::$option{table_name};
92			use strict;
93	wakaba	1.5	our \$VERSION = q(@{[sprintf '%04d.%02d%02d', (gmtime)[5]+1900, (gmtime)[4]+1, (gmtime)[3]]});
94	wakaba	1.1
95			## These tables are embeded in binary, so that your editor
96			## might break the data or might hang up.
97
98
99
100
101
102
103
104
105
106			#
107
108	wakaba	1.5	our %L2U = map {Encode::_utf8_on (\$_) if length \$_ > 1; \$_} unpack
109	wakaba	1.2	(q{$pack}, <<'END');
110	wakaba	1.1	EOH
111
112			print $tbl,"\nEND\n";
113
114			print <<EOH;
115	wakaba	1.5	our %U2L = reverse %L2U;
116	wakaba	1.1	sub import {
117	wakaba	1.2	\$Encode::Table::TABLE{$option{table_name}_to_ucs} = \\%L2U;
118			\$Encode::Table::TABLE{ucs_to_$option{table_name}} = \\%U2L;
119	wakaba	1.1	}
120			1;
121	wakaba	1.2	### $option{table_name}.pm ends here
122	wakaba	1.1	EOH
123
124	wakaba	1.2
125			=head1 SEE ALSO
126
127			Encode::Table, internal.pl
128
129			=head1 AUTHOR
130
131			Nanashi-san
132
133			=head1 LICENSE
134
135			Copyright 2002 AUTHOR
136
137			This library is free software; you can redistribute it
138			and/or modify it under the same terms as Perl itself.
139
140			AUTHOR does NOT claim any right to the data generated by
141			this script. License of generated data fully depends
142			author of source data.
143
144			=cut
145
146	wakaba	1.5	1; ## $Date: 2002/10/12 07:27:01 $
147	wakaba	1.1	### tbl2pm.pl ends here