pbot/misc/wordmorph/wordmorph-mkdb

#!/usr/bin/perl

use warnings;
use strict;

use Storable;
use Text::Levenshtein::XS 'distance';

die "usage: $0 <wordlist file>\n" unless @ARGV == 1;

#thanks to Limbic~Region, http://perlmonks.org/index.pl?node_id=180961
open (my $wordlist, '<', $ARGV[0]) or die "Unable to open `$ARGV[0]` for reading: $!\n";
my ($db, %data);

while (<$wordlist>) {
    chomp;
    next if $_ !~ /^[a-z]{3,7}$/;
    push @{$data{length()}}, $_;
}

for my $len (keys %data) {
    my $end = $#{$data{$len}};
    for my $i (0 .. $end - 1) {
        my $word = $data{$len}[$i];
        for my $j ($i + 1 .. $end) {
            my $test = $data{$len}[$j];
            if (distance($word, $test) == 1) {
                push @{$db->{$len}{$word}}, $test;
                push @{$db->{$len}{$test}}, $word;
            }
        }
    }
}

store $db, 'wordmorph.db';