diff options
Diffstat (limited to 'lib/unicore/mktables')
| -rw-r--r-- | lib/unicore/mktables | 56 |
1 files changed, 51 insertions, 5 deletions
diff --git a/lib/unicore/mktables b/lib/unicore/mktables index 060a0e63e0..3328f69326 100644 --- a/lib/unicore/mktables +++ b/lib/unicore/mktables @@ -103,6 +103,9 @@ my %In; my $InId = 0; my %InIn; +my %InScript; +my %InBlock; + # # Read in the Unicode.txt, the main Unicode database. # @@ -355,8 +358,9 @@ for my $script (sort { $a->[0] <=> $b->[0] } @Scripts) { extend($Script{$name}, $last); } unless (defined $In{$name}) { - $In{$name} = $InId++; - $InIn{$name} = $Script{$name}; + $InScript{$InId} = $name; + $In{$name} = $InId++; + $InIn{$name} = $Script{$name}; } } @@ -382,11 +386,19 @@ if (open(my $Blocks, "Blocks.txt")) { next unless /^([0-9A-Fa-f]+)\.\.([0-9A-Fa-f]+)\s*;\s*(.+?)\s*$/; my ($first, $last, $name) = ($1, $2, $3); + my $origname = $name; # If there's a naming conflict (the script names are # in uppercase), the name of the block has " Block" # appended to it. - $name = "$name Block" if defined $In{"\U$name"}; + my $pat = $name; + $pat =~ s/([- _])/(?:[-_]|\\s+)?/g; + for my $i (values %InScript) { + if ($i =~ /^$pat$/i) { + $name .= " Block"; + last; + } + } append(\@Blocks, $first, $name); append($Blocks{$name} ||= [], $first, $name); @@ -395,8 +407,9 @@ if (open(my $Blocks, "Blocks.txt")) { extend($Blocks{$name}, $last); } unless (defined $In{$name}) { - $In{$name} = $InId++; - $InIn{$name} = $Blocks{$name}; + $InBlock{$InId} = $origname; + $In{$name} = $InId++; + $InIn{$name} = $Blocks{$name}; } } } else { @@ -591,6 +604,39 @@ EOT mapping(\%In, "In"); +# +# Append the InScript and InBlock mappings. +# These are needed only if Script= and Block= syntaxes are used. +# + +if (open(my $In, ">>In.pl")) { + print $In <<EOT; + +%utf8::InScript = +( +EOT + for my $i (sort { $a <=> $b } keys %InScript) { + printf $In "%4d => '$InScript{$i}',\n", $i; + } + print $In <<EOT; +); +EOT + + print $In <<EOT; + +%utf8::InBlock = +( +EOT + for my $i (sort { $a <=> $b } keys %InBlock) { + printf $In "%4d => '$InBlock{$i}',\n", $i; + } + print $In <<EOT; +); +EOT +} else { + die "$0: In.pl: $!\n"; +} + # Easy low-calorie cheat. use File::Copy; copy("In/$In{Noncharacter_Code_Point}.pl", "Is/Cn.pl"); |