polog

ファイル

rake db:migrateで

    create_table :chars do |t|

      t.column :char, :string

      t.column :byte, :integer

    end

    add_index :chars, :char

    add_index :chars, :byte

    create_table :similarities do |t|

      t.column :char1_byte, :integer

      t.column :char2_byte, :integer

      t.column :similarity, :integer

    end

    add_index :similarities, [:char1_byte, :similarity]

    add_index :similarities, :similarity

#!/usr/bin/env /path/to/yours/kanji_rails/script/runner

$KCODE = "u"

require 'jcode'

require 'RMagick'


def diff_a(a1, a2)

  result = 0

  (0..399).to_a.each{|i| result +=1 if a1[i] == a2[i]}

  result

end

chars_a = []

open("kanji.txt").each_line do |line|

  line.chomp

  chars_a << line.split(/\s/)

end

chars_a.flatten!

puts "chars retrieved"

chars_h = Hash.new {|h, k| h[k] = {}}

chars_q = []

chars_a.each do |char|

  i = eval("0x#{char.unpack('C*').map{|i| i.to_s(16)}.join}")

  chars_h[char][:byte] = i

  chars_q << {

    :byte => i,

    :char => char

  }

end

puts "chars analyzed"

chars_a = chars_a.sort_by{|c| chars_h[c][:byte]}

Char.create(chars_q)

puts "chars created"

pos = 0

chars_a.each do |char|

  img = Magick::ImageList.new()

  img.new_image(20,20)

  text = Magick::Draw.new

  text.gravity(Magick::CenterGravity)

  text.stroke('transparent')

  text.fill("black")

  text.font("mincho.ttf")

  text.font_size(20)

  text.text(0,0,char)

  text.draw(img)

  

  px_a = []

  20.times do |x|

    20.times do |y|

      px_a << (img.pixel_color(x,y).intensity.to_i != 255)

    end

  end

  chars_h[char][:px_a] = px_a

  pos += 1

  puts "#{pos} chars pixel analyzed" if pos % 100 == 0

end

puts "chars pixel analyzed"

pos = 0

similarity_q = []

chars_a.each do |char1|

  chars_a.each do |char2|

    next if char1 == char2

    similarity_q << {

      :char1_byte => chars_h[char1][:byte],

      :char2_byte => chars_h[char2][:byte],

      :similarity => diff_a(chars_h[char1][:px_a], chars_h[char2][:px_a])

    }

    pos += 1

    if pos % 1000 == 0

      Similarities.create(similarity_q)

      similarity_q = []

      puts "#{pos} char pairs analyzed"

    end

  end

end

クリックで全体表示