CC-CEDICT loader

cedictloader.rb

class CedictLoader
  include Enumerable
  
  URL = ENV['CEDICT'] || 'http://www.mdbg.net/chindict/export/cedict/cedict_1_0_ts_utf-8_mdbg.txt.gz'

  def initialize(input = nil)
    @input = input || (
      require 'zlib'
      require 'open-uri'
      Zlib::GzipReader.new(open(URL)))
    @headers = []
    @input.lines.each do |line|
      if line =~ /^# /
        @headers << line
      else
        break
      end
    end
  end
  
  def process_input(line)
    if line.strip =~ /^([^\s]*) ([^\s]*) \[([\w\d: ]+)\](.*)/
      [$1,$2,$3,$4].map{|x| x.strip}
    else
      line
    end
  end

  def each
    if block_given?
      @input.lines.each do |line|
        yield process_input(line) if line !~ /^# /
      end
    else
      enum_for(:each)
    end
  end
end