From: Martin DeMello <martindemello@...>
Date: 2007-01-11T23:25:32+09:00
Subject: Re: [QUIZ] Word Blender (#108)

On 1/11/07, Fedor Labounko <fedor.labounko@gmail.com> wrote:
> On 1/11/07, Martin DeMello <martindemello@gmail.com> wrote:
> >
> > This just solves the find-all-subwords problem:
> >
> > target = ARGV[0]
> > dict = ARGV[1] || 'sowpods'
> >
> > reduced = target.split(//).sort.uniq.join
> > primes = [2, 3, 5, 7, 11, 13]
> > factors = []
> > reduced.split(//).each_with_index {|e, i|
> >   factors[e[0]] = primes[i]
> > }
> >
> > target_num = 1
> > target.each_byte {|i| target_num *= factors[i]}
> >
> > IO.foreach(dict) {|word|
> >   word.chomp!
> >   next unless (word =~ /^[#{reduced}]+$/) &&
> >     (word.length < 7) && (word.length > 2)
> >   p = 1
> >   word.each_byte {|i| p *= factors[i]}
> >   puts word if target_num % p == 0
> > }
>
> That's neat, and is a good general way of checking for inclusion (so you can
> extend it past characters in a string which might not have an .include?
> method). You probably don't want that .uniq in there though as that excludes
> you from matching 'hell' out of 'hello', for example.

No, I do need the uniq since I want one prime per unique letter in the
target word. I then calculate its signature by a loop over the entire
word, not the reduced word, so "hell" is indeed matched for "hello"
but not for, say, "whelps".

martin