charset: split unicode includes from static encodings

[sheet.git] / charset.inc.pl

diff --git a/charset.inc.pl b/charset.inc.pl
new file mode 100644 (file)
index 0000000..020cb2d
--- /dev/null
+++ b/charset.inc.pl
@@ -0,0 +1,59 @@
+use 5.014;
+use warnings;
+use utf8;
+
++{
+       %{ Data('./charset-encoding') },
+
+       ''             => {setup => sub {
+               my $row = shift;
+               $row->{offset} = delete $row->{startpoint};
+               $row->{set} = 'Unicode characters';
+               my $block = $row->{offset} >> 8;
+               $row->{endpoint} ||= ($block + 1 << 8) - 1;
+               $block == ($row->{endpoint} >> 8) or undef $block;
+
+               $row->{table} = join '', map { chr =~ s/\A\p{Unassigned}\z/�/r }
+                       $row->{offset} .. $row->{endpoint};
+               utf8::upgrade($row->{table});  # prevent latin1 output
+
+               $row->{endpoint} -= $row->{offset};
+
+               if (defined $block) {
+                       $row->{set} = sprintf 'Unicode block U+%02Xxx', $block;
+                       $row->{offset} %= 0x100;
+               }
+
+               return $row;
+       }},
+       u              => {setup => sub {
+               my $row = shift;
+               state $celldata = eval { Data('charset-unicode') }
+                       or Alert('Table data could not be read', ref $@ && $@->[1]);
+               $row->{cell} = $celldata;
+
+               $row->{endpoint} ||= 0x1FFF;
+               $row->{set} = 'Unicode ' . (
+                       $row->{startpoint} <  0x1000 && $row->{endpoint} < 0x1000 ? 'BMP' :
+                       $row->{startpoint} >= 0x1000 && $row->{endpoint} < 0x2000 ? 'SMP' :
+                       'allocations'
+               );
+               return $row;
+       }},
+       uu             => {setup => sub {
+               my $row = shift;
+               $row->{cell} = eval { Data('charset-ucplanes') }
+                       or Alert('Table data could not be read', ref $@ && $@->[1]);
+               $row->{endpoint} ||= 0x3FF;
+               $row->{set} = 'Unicode planes';
+               return $row;
+       }},
+       utf8           => {setup => sub {
+               my $row = shift;
+               $row->{set} = 'UTF-8';
+               $row->{cell} = eval { Data('charset-utf8') }
+                       or Alert('Table data could not be read', ref $@ && $@->[1]);
+               return $row;
+       }},
+       'utf-8'        => 'utf8',
+};