Romanization Rule into Trie

kanatrie.rb

#!/usr/bin/env ruby

# kanatrie.rb : ローマ字の綴りをトライ木に(ヘボン式・訓令式・日本式に対応)
#   by takehikom

# required: KAKASI, Graphviz

def create_romaji_kana_hash(option = {})
  if option[:label]
    puts "==== #{option[:label]} ===="
  end

  str1 = "あいうえおかきくけこさしすせそたちつてとなにぬねのはひふへほまみむめもやゆよらりるれろわをんがぎぐげござじずぜぞだぢづでどばびぶべぼぱぴぷぺぽ"
  str2 = "きしちにひみりぎじぢびぴ"
  open("kana.txt", "w") do |f_out|
    f_out.puts str1.split(//)
    f_out.puts str2.split(//).map {|c| [c + "ゃ", c + "ゅ", c + "ょ"]}
  end

  command = "kakasi -Ha -r -iutf8 <kana.txt >romaji.txt"
  command.sub!(/-r/, "-rkunrei") if option[:kunrei]
  puts command
  system command

  a1 = open("kana.txt").read.split(/\n/)
  a2 = open("romaji.txt").read.split(/\n/)
  r2c_h = {}
  c2r_h = {}
  a1.each_with_index do |c1, i|
    c2 = a2[i]
    if option[:kunrei] && !option[:di]
      if c1 == "ぢ"
        c2 = "zi"
      elsif c1 == "づ"
        c2 = "zu"
      end
    end
    if !option[:wo] && c1 == "を"
      c2 = "o"
    end
    puts "#{c2} => #{c1}"
    if r2c_h.key?(c2)
      r2c_h[c2] = r2c_h[c2] + "," + c1
      puts "(#{c2} => #{r2c_h[c2]})"
    else
      r2c_h[c2] = c1
    end
    c2r_h[c1] = c2
  end

  File.unlink("kana.txt")
  File.unlink("romaji.txt")

  [r2c_h, c2r_h]
end

rk_h = {}
rk_h[:hepburn] = create_romaji_kana_hash(:ji => true, :label => "hepburn")
rk_h[:kunrei] = create_romaji_kana_hash(:kunrei => true, :label => "kunrei")
rk_h[:nippon] = create_romaji_kana_hash(:kunrei => true, :di => true, :wo => true, :label => "nippon")

# exit

def generate_trie(c2r_h, basename, label)
  puts "start: #{label}"

  open("#{basename}.dot", "w") do |f_out|
    f_out.puts "digraph #{basename} {"
    f_out.puts "  graph [rankdir = LR, label = \"#{label}\"];"
    node_a = []
    node_a << "empty [label = \"\"];"
    path_a = []
    romend_a = []

    c_keys = c2r_h.keys.sort_by {|key|
      case key[0]
      when /[あいうえおかきくけこさしすせそたちつてとなにぬねのはひふへほまみむめもやゆよらりるれろわをん]/
        "1" + key
      when /[がぎぐげござじずぜぞだぢづでどばびぶべぼ]/
        "2" + key
      when /[ぱぴぷぺぽ]/
        "3" + key
      else
        "9" + key
      end
    }

    c_keys.each do |c|
      rom = c2r_h[c]
      romseq_a = (1..(rom.length)).to_a.map {|i| rom[0, i]}
      romend = "x_" + c.unpack("H*").first
      romend_a << romend
      node_a << "#{romend} [shape = box, label = \"#{c}\"];"
      path_a << "empty -> #{romseq_a.first};"
      0.upto(romseq_a.length - 2) do |i|
        path_a << [romseq_a[i], romseq_a[i + 1]].join(" -> ") + ";"
      end
      path_a << "#{rom} -> #{romend} [dir = none, style = dotted];"
    end

    node_a.uniq!
    path_a.uniq!

    f_out.puts
    f_out.puts node_a.map {|item| "  " + item}
    f_out.puts
    f_out.puts path_a.map {|item| "  " + item}

    f_out.puts
    f_out.puts "  {rank = same; #{romend_a.join('; ')}}"

    f_out.puts "}"
  end

  command = "dot -Tpng #{basename}.dot -o #{basename}.png"
  puts command
  system command
end

label = "ヘボン式ローマ字のトライ木"
basename = "kanatrie_h"
generate_trie(rk_h[:hepburn][1], basename, label)

label = "訓令式ローマ字のトライ木"
basename = "kanatrie_k"
generate_trie(rk_h[:kunrei][1], basename, label)

label = "日本式ローマ字のトライ木"
basename = "kanatrie_n"
generate_trie(rk_h[:nippon][1], basename, label)