#! /usr/bin/ruby
# -*- coding: utf-8 -*-

BASEDIR = ARGV.length > 0 ? ARGV[0] : "."

if !File::directory?(BASEDIR)
  printf("%s is not a directory\n", BASEDIR)
  exit(1)
end

pattern = sprintf("%s/*.TXT", BASEDIR)
Dir.glob(pattern) do |path|
  dictname = path.sub(/.*\//, "")
  dictname = dictname.sub(/[-_0-9]*\.TXT$/, "").downcase
  File::open(path) do |infile|
    outpath = sprintf("eijiro-%s.tsv", dictname)
    File::open(outpath, "w") do |outfile|
      seq = 0
      infile.each do |line|
        begin
          line = line.encode('UTF-8', 'Windows-31J')
        rescue => e
          p e
          next
        end
        line = line.gsub(/\s+/, " ")
        line = line.strip
        line = line.sub(/^■/, "")
        face = line.sub(/ +:.*/, "")
        text = line.sub(/[^:]* +: +/, "")
        part = ""
        if face.match(/ *{[^}]+}$/)
          part = face.sub(/.*{([^}]+)}$/, '\1')
          face = face.sub(/ *{[^}]+}$/, "")
        end
        key = face.downcase
        text = text.gsub(/｛[^｝]+｝/, "")
        text = text.sub(/ *◆([a-z]+:|【URL】|【出典】).*$/, "")
        seq += 1
        printf(outfile, "%s\t%d\t%s\t%s\t%s\n", key, seq, face, part, text)
        printf("%s: %s: %d records done\n", $0, dictname, seq) if seq % 1000 == 0
      end
    end
  end
end