英文テキストファイルの各アルファベットの数を出力して割合をグラフで表示する

lettercount.rb

require "gnuplot"
# (gem install gnuplot)
# (brew install gnuplot)

INPUT_DATA = ARGV[0] || raise("input file name is not given")
OUTPUT_DATA = ARGV[1] || "output.txt"
OUTPUT_GRAPH = ARGV[2] || "output.eps"

n = 0

hash = Hash.new(0) #デフォルトの値を設定(登場していないのは0)
File::open(INPUT_DATA) {|file| 
  file.each {|line| 
  	line = line.downcase.split(//).delete_if{|c| !('a'..'z').include?(c) }
  	while line.length > 0
  	  hash[line[0]] = line.count(line[0]) 
  	  line.delete(line[0])
  	  n += 1
  	end
  }
}

letters_count = hash.values.reduce {|sum, n| sum + n }
hash_sorted = hash.sort {|(k1, v1), (k2, v2)| v2 <=> v1 } #[[k1, v1], [k2, v2]]みたいな形式になるから
hash_sorted = Hash[*hash_sorted.flatten] #[k1, v1, k2, v2 ...]にしてハッシュ化
File::open(OUTPUT_DATA, "w") {|file|
	file.puts(letters_count)
	hash_sorted.each {|key, value| file.puts("#{key},#{value}")}
}

Gnuplot.open do |gp|
	Gnuplot::Plot.new( gp ) do |plot|
        # set image paramter
        plot.terminal 'postscript eps enhanced color'
        tmp = Array.new(('a'..'z').to_a.length) { |i| "'#{('a'.ord + i).chr}' #{i + 1}" }
        plot.set 'xtics' , "(" + tmp.join(', ') + ")"
        plot.set 'output', OUTPUT_GRAPH
        plot.set 'xrange', "[0:#{('a'..'z').to_a.length + 1}]"
        plot.set 'yrange', "[0:0.15]"
        plot.set 'key', 'outside' #凡例項目の表示をグラフ外にする
        # set title and labels
	plot.ylabel 'percentage(%)'
	plot.xlabel 'letter'
	plot.title  'letter frequency'

        def formatData(hash, letters_count)
            x = (1..('a'..'z').to_a.length).to_a
            y = Array.new(('a'..'z').to_a.length) { |i| hash[('a'.ord + i).chr] / letters_count.to_f }
            return [x, y]
        end

        # make fill lines with title
        def makeBoxGraph( data , title )
            Gnuplot::DataSet.new( data ) do |ds|
                ds.title = title 
                ds.with = "boxes"
            end
        end

        # plot graph
        plot.data << makeBoxGraph(formatData(hash, letters_count) , ARGV[0])
	end
end

`open #{OUTPUT_GRAPH}`