charset.inc.pl

   1 use 5.014;
   2 use warnings;
   3 use utf8;
   4
   5 +{
   6         %{ Data('./charset-encoding') },
   7
   8         ''             => {setup => sub {
   9                 my $row = shift;
  10                 $row->{offset} = delete $row->{startpoint};
  11                 $row->{set} = 'Unicode characters';
  12                 my $block = $row->{offset} >> 8;
  13                 $row->{endpoint} ||= ($block + 1 << 8) - 1;
  14                 $block == ($row->{endpoint} >> 8) or undef $block;
  15
  16                 $row->{table} = join '', map { chr =~ s/\A\p{Unassigned}\z/�/r }
  17                         $row->{offset} .. $row->{endpoint};
  18                 utf8::upgrade($row->{table});  # prevent latin1 output
  19
  20                 $row->{endpoint} -= $row->{offset};
  21
  22                 if (defined $block) {
  23                         $row->{set} = sprintf 'Unicode block U+%02Xxx', $block;
  24                         $row->{offset} %= 0x100;
  25                 }
  26
  27                 return $row;
  28         }},
  29         u              => {setup => sub {
  30                 my $row = shift;
  31                 state $celldata = eval { Data('charset-unicode') }
  32                         or Alert('Table data could not be read', ref $@ && $@->[1]);
  33                 $row->{cell} = $celldata;
  34
  35                 $row->{endpoint} ||= 0x1FFF;
  36                 $row->{set} = 'Unicode ' . (
  37                         $row->{startpoint} <  0x1000 && $row->{endpoint} < 0x1000 ? 'BMP' :
  38                         $row->{startpoint} >= 0x1000 && $row->{endpoint} < 0x2000 ? 'SMP' :
  39                         'allocations'
  40                 );
  41                 return $row;
  42         }},
  43         uu             => {setup => sub {
  44                 my $row = shift;
  45                 $row->{cell} = eval { Data('charset-ucplanes') }
  46                         or Alert('Table data could not be read', ref $@ && $@->[1]);
  47                 $row->{endpoint} ||= 0x3FF;
  48                 $row->{set} = 'Unicode planes';
  49                 return $row;
  50         }},
  51         utf8           => {setup => sub {
  52                 my $row = shift;
  53                 $row->{set} = 'UTF-8';
  54                 $row->{cell} = eval { Data('charset-utf8') }
  55                         or Alert('Table data could not be read', ref $@ && $@->[1]);
  56                 return $row;
  57         }},
  58         'utf-8'        => 'utf8',
  59 };