mecabのユーザー辞書を作成する書き捨てスクリプト

userdic_source_generator.rb

require "csv"

class String
  # 参考: http://d.hatena.ne.jp/hirokan55/20100210/p1
  def cost
    [-36000, -400 * (self.length ** 1.5)].max.to_i
  end
end

class UserdicSourceGenerator
  def self.from(source_path)
    new(source_path)
  end

  def initialize(source_path)
    @source = CSV.table(source_path)
  end

  def to(output_path, opts = {})
    CSV.open(output_path, "wb") do |csv|
      @source.each do |source_row|
        row = []
        row << source_row[0]       # 表層形
        row << 0                   # 左文脈ID
        row << 0                   # 右文脈ID
        row << source_row[0].cost  # コスト
        row << "名詞"              # 品詞
        row << "固有名詞"          # 品詞細分類1
        row << "*"                 # 品詞細分類2
        row << "*"                 # 品詞細分類3
        row << "*"                 # 活用形
        row << "*"                 # 活用型
        row << source_row[1]       # 原形
        row << "*"                 # 読み
        row << "*"                 # 発音
        row << opts[:tag]          # 登録タグ

        csv << row
      end
    end
  end
end

UserdicSourceGenerator.from("input.csv").to("output.csv", tag: "my_dictionary")