Given lots of tags, keep a running count of the top-n

top_n_tags.rb

require "pp"

class TagManager
  def initialize(top_tags_to_keep)
    @all_tags = Hash.new(0)
    @top_tags = []
    @top_tags_to_keep = top_tags_to_keep
  end
  
  def search(tag)
    @all_tags[tag] = @all_tags[tag] + 1
    recalculate_top_tags(tag, @all_tags[tag])
  end
  
  def add_to_top_tags(tag)
    @top_tags << tag unless @top_tags.include? tag
  end

  def recalculate_top_tags(tag, hits)
    
    if @top_tags.size < @top_tags_to_keep
      add_to_top_tags(tag)
    end
  
    if @all_tags[@top_tags.last] < hits
      add_to_top_tags(tag)
    end

    @top_tags.sort! do |a,b|
      @all_tags[b] <=> @all_tags[a]
    end
    @top_tags = @top_tags[0..(@top_tags_to_keep - 1)]
  end

end

mgr = TagManager.new(5)

tags = %w(algorithm code test python ruby java c# .Net Perl jQuery html design-patterns c C++ hashkell Objective-C List Scala Clojure)
tags.each do |tag|
  rand(10000).times { |n| mgr.search(tag) }
end

pp mgr


__END__

Output:
#<TagManager:0x10036e420
 @all_tags=
  {"html"=>5921,
   "c#"=>5681,
   "code"=>6040,
   "C++"=>154,
   "c"=>9837,
   "java"=>8012,
   "algorithm"=>4199,
   "Clojure"=>885,
   "Scala"=>3243,
   "Objective-C"=>4394,
   "Perl"=>4316,
   "python"=>3333,
   "List"=>9305,
   "jQuery"=>8049,
   "ruby"=>7665,
   "design-patterns"=>565,
   "test"=>9943,
   "hashkell"=>5033,
   ".Net"=>2606},
 @top_tags=["test", "c", "List", "jQuery", "java"],
 @top_tags_to_keep=5>