gen-combining.pl

use strict;
use warnings;

my $range_start = undef;
my $count = 0;

print "\tstatic const struct mbinterval combining[] = {";

foreach my $line (<ARGV>)
{
    chomp $line;
    my @fields = split ';', $line;
    my $codepoint = hex $fields[0];

    next if $codepoint > 0xFFFF;

    if ($fields[2] eq 'Me' || $fields[2] eq 'Mn')
    {
        # combining character, save for start of range
        if (!defined($range_start))
        {
            $range_start = $codepoint;
        }
    }
    else
    {
        # not a combining character, print out previous range if any
        if (defined($range_start))
        {
            if ($count++ % 3 == 0)
            {
                print "\n\t\t";
            }
            else
            {
                print " ";
            }
            printf "{0x%04X, 0x%04X},", $range_start, $codepoint;
            $range_start = undef;
        }
    }
}

print "\n\t};\n";