prime 1.0.0.1-2 / usr / bin / prime-dict-convert

#!/usr/bin/env ruby
# prime-dict-convert: A dictionary converter from a dictionary of another IME
# $Id: prime-dict-convert.src,v 1.4 2005/03/07 07:51:34 komatsu Exp $
#
# Copyright (C) 2003 Hiroyuki Komatsu <komatsu@taiyaki.org>
#     All rights reserved.
#     This is free software with ABSOLUTELY NO WARRANTY.
#
# You can redistribute it and/or modify it under the terms of 
# the GNU General Public License version 2.

PRIME_LIBDIR = '/usr/lib/ruby/1.8'
$LOAD_PATH.unshift(PRIME_LIBDIR) unless $LOAD_PATH.member?(PRIME_LIBDIR)

require 'prime/taiyaki'
require 'getoptlong'
require 'prime/makedict/dictformat-skkdic'
require 'prime/makedict/dictformat-pubdic'
require 'prime/makedict/dictformat-cannadic'
require 'prime/makedict/dictformat-adambnc'
require 'prime/makedict/dictformat-wordnet'
require 'prime/makedict/dictformat-usage'
require 'prime/makedict/basicdict'

begin
  ## The following files require MeCab.
  Kernel::require('prime/makedict/dictformat-wordlist')
  Kernel::require('prime/makedict/dictformat-text')
rescue LoadError
  FLAG_MECAB = false
else
  FLAG_MECAB = true
end

class PrimeDictConvertCommand
  include Debug

  def initialize (command_name)
    @debug_mode = false
    @is_interactive = true

    @indexing   = true
    @conversion = true
    @output_dictname = nil
    @input_dictnames = []

    @formats = {:prime    => [DictFormat],
                :pubdic   => [DictFormatPubdic],
                :canna    => [DictFormatCannadic],
                :skk      => [DictFormatSkkdic],
                :usage    => [DictFormatUsage],
                :adambnc  => [DictFormatAdamBNC],
                :wordnet  => [DictFormatWordNet],
    }
    if FLAG_MECAB then
      @formats[:wordlist] = DictFormatWordlist
      @formats[:text]     = DictFormatText
      @formats[:gaim]     = DictFormatGaim
    end

    @command_name = command_name
    @version = '1.0.0.1'
    @usage = <<"EOF"
#{@command_name}:#{@version} -- converts from a dictionary of another IME.

  Usage: #{@command_name} output_dictname --<format> input_dictname [options]
       --append    : append words to the existent output dictionary. (default)
       --overwrite : delete the existent output dictionary and make new dict.

       --prime    :  merge the input_dict with the output_dict
       --canna    :  convert from the canna dict
       --skk      :  convert from the skk dict
       --pubdic   :  convert from the pubdic dict
       --usage    :  convert from the usage dict (beta)
       --adambnc  :  convert from the Adam\'s BNC word frequency list (beta)
       --wordnet  :  convert from the WordNet cntlist
       --wordlist :  convert from the wordlist without the readings and POSes
                     (beta)
       --text     :  convert from the plain text file (beta)
       --gaim     :  convert from the Gaim log (beta)

       --no-index :  convert only without indexing
  -q,  --quiet    :  run this command and show nothing.
  -v,  --version  :  show the version and exit
  -h,  --help     :  show this help and exit
  -d,  --debug    :  run under debug mode
EOF
  end

  def main ()  
    parse_options()

    if @output_dictname.nil? or @input_dictnames.empty? then
      print_usage()
      exit()
    end
    convert_dicts()
    index_dict()
  end

  def convert_dicts
    basedict = DictFormat.new(@is_interactive)
    if @flag_append then
      basedict.load_existent_dict(@output_dictname)
      dict = basedict.dict
    end
    dict = basedict.dict

    converter = nil
    @input_dictnames.each {|(type, filename)|
      if @formats.has_key?(type) then
        converter = @formats[type][0].new(@is_interactive)
      else
	puts "Invalid format option..."
	exit()
      end

      converter.dict = dict
      converter.load_external_dict(filename)
      dict = converter.dict
    }
    converter.save_dict(@output_dictname)
  end

  def index_dict ()
    if @flag_index == false then
      return
    end
    indexer = PrimeBasicdict.new(@output_dictname, @is_interactive)
    indexer.make_pos_table()
    indexer.make_basicdict_indexes()
  end

  def print_version ()
    puts "#{@command_name}:#{@version}"
  end
  
  def print_usage ()
    puts @usage
  end

  private
  def parse_options ()
    options = {}
    parser = GetoptLong.new()
    option_list = [
      ['--help',       '-h',   GetoptLong::NO_ARGUMENT],
      ['--version',    '-v',   GetoptLong::NO_ARGUMENT],
      ['--debug',      '-d',   GetoptLong::NO_ARGUMENT],
      ['--quiet',      '-q',   GetoptLong::NO_ARGUMENT],
      ['--append',             GetoptLong::NO_ARGUMENT],
      ['--overwrite',          GetoptLong::NO_ARGUMENT],
      ['--no-index',           GetoptLong::NO_ARGUMENT],
    ]
    @formats.keys.each { | key |
      option_list.push( ['--' + key.to_s(), GetoptLong::REQUIRED_ARGUMENT] )
    }
    parser.set_options(*option_list)

    parser.each_option {|option, arg|
      options[option.sub(/^--/, '')] = arg
    }

    if options['version'] then
      print_version()
      exit()
    elsif options['help'] then
      print_usage()
      exit()
    end

    if options['debug'] then
      $DEBUG = true
    end
    if options['quiet'] then
      @is_interactive = false
    end

    @output_dictname = ARGV[0]

    @flag_append = true
    if options['append'] then
      @flag_append = true
    elsif options['overwrite'] then
      @flag_append = false
    end

    if options['no-index'] then
      @flag_index = false
    else
      @flag_index = true
    end

    @formats.keys.each { | format |
      argument = options[format.to_s]
      if argument then
        @input_dictnames.push([format, argument])
      end
    }
        
  end
end



if File::expand_path($0) == File::expand_path(__FILE__) then
  conv_command = PrimeDictConvertCommand.new(File::basename($0))
  conv_command.main()
end


# --- Text properties for Emacs. ---
# Local variables:
# mode: ruby
# End: