[ruby-talk:8160] Re: speedup of anagram finder

David Alan Black <[e-mail]> writes:

> For what it's worth, here's a tweaked/hacked version of your code.
> Some of the changes are things mentioned above; others are just little
> experiments, some of which didn't seem to matter much for speed.

I did a little tweaking too. It seemed to me that a lot of effort was
going to the the final output loop, testing to see if an entry had two 
or more words in it, where perhaps only 1% of them do. So, I added
another hash which records the keys of known anagrams:


     def fa6(words, out = STDOUT)
       anagrams = {}
       keys = {}
       word, key = nil

       for word in words do
         word.chomp!
         word.downcase!
         key = word.split('')
         key.sort!
         key = key.join
         if anagrams[key]
           anagrams[key] << word
           keys[key] = 1
         else
           anagrams[key] = [ word ]
         end
       end
       for key in keys.keys
         out.puts anagrams[key].join(' ')
       end
     end

This benchmarks slightly faster than David's latest:

     dave[~/tmp 0:25:05] head -20000 /usr/dict/words | ruby -w anag.rb
                           user     system      total        real
     original          5.060000   0.090000   5.150000 (  5.149464)
     omit to_s         2.700000   0.000000   2.700000 (  2.709957)
     block var         2.230000   0.010000   2.240000 (  2.238881)
     lst[1]            2.220000   0.010000   2.230000 (  2.247656)
     sort!             2.140000   0.010000   2.150000 (  2.165254)
     david             1.540000   0.000000   1.540000 (  1.541778)
     keys              1.400000   0.000000   1.400000 (  1.398975)


Next!


Dave