saykana

saykana.rb

#!/usr/bin/env ruby
# vim:fileencoding=utf-8

require 'MeCab'
require 'kakasi'
require 'kconv'
require 'uri'

# http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-list/33639
class String
  def without_match(re)
    re = Regexp.new('(' + re.source + ')')
    split(re).map { |s|
      if re =~ s
        s
      else
        yield s
      end
    }.join
  end

  def remove_url
    self.gsub(URI.regexp(['http','https']),'')
  end
end

def kakasi(word)
  return Kakasi::kakasi("-JK -HK",word.tosjis).toutf8
end

@chasen = MeCab::Tagger.new()
def pronounce(word)
  return @chasen.parse(word).map{|l|l.chomp}.reject{|l|l=="EOS"}.map { |line|
    word, data = line.split("\t")
    #puts
    #puts "word : #{word}"
    # 品詞, 品詞細分類1, 品詞細分類2, 品詞細分類3, 活用形, 活用型, 原型, 読み, 発音
    speech, category1, category2, category3, inflected_form, inflected_format, original, reading, pronounciation = data.split(",") unless data.nil?
    #puts "data : #{data}"
    #puts "read : #{pronounciation unless pronounciation.nil?}"
    pronounciation.nil? ? kakasi(word) : pronounciation
  }
end

def sayKana(str)
  kana = str.split(" ").map{|word|
    word.without_match(/\d[\d.-]*(?:年|月|日|時|分|秒|円|階|ヶ月|カロリー|級|行|曲|キロ|件|個
|人|才|時間|台|丁目|月|番|本|匹|％)?/) {|word|
      pronounce(word)
    }
  }.join(" ").remove_url

  puts kana

  fork do
    open('|SayKana -s 87 -f -', 'w') do |f|
      f.puts kana
    end
  end
end

if __FILE__ == $0
  s = ARGV[0]
  puts s
  sayKana(s)
end