From: Gavin Kistner <gavin.kistner@...>
Date: 2006-10-13T05:08:31+09:00
Subject: Re: Isolating non-unique items in an array

From: Gavin Kistner 
> But of course, other posts in this thread are likely faster.

How much faster? Much.

class Array
  def dcount1
    Hash[ *uniq.map{ |e|
      if ( count = grep(e).size ) > 1
        [e, count]
      end
    }.compact.flatten ]
  end
  
  def dcount2
    inject( Hash.new(0) ){ |counts,e|
      counts[e]+=1
      counts
    }.reject{ |e,count|
      count==1
    }
  end
  
  def dcount3
    counts = Hash.new(0)
    each{ |e| counts[e]+=1 }
    counts.reject{ |e,count| count==1 }
  end
  
end

a = %w| a a a a b c d d e b c f g h h i j k l l m n o p a b f h l m a a
c d d b h l l m h |

runs = 10000
require 'benchmark'
Benchmark.bmbm{ |x|
  x.report( 'dcount1' ){
    runs.times{ a.dcount1 }
  }
  x.report( 'dcount2' ){
    runs.times{ a.dcount2 }
  }
  x.report( 'dcount3' ){
    runs.times{ a.dcount3 }
  }
}

#=>               user     system      total        real
#=> dcount1   5.641000   0.000000   5.641000 (  5.641000)
#=> dcount2   2.000000   0.015000   2.015000 (  2.016000)
#=> dcount3   1.360000   0.016000   1.376000 (  1.375000)