chise/character.rb

   1 # Copyright (C) 2002-2004 Kouichirou Eto, All rights reserved.
   2
   3 require "singleton"
   4 require "chise/parser"
   5 require "chise/chisedb"
   6 require "chise/iconv"
   7 require "chise/utf8"
   8 require "chise/ids"
   9
  10 module CHISE
  11   class CharacterFactory # generate Character object and cache them
  12     include Singleton
  13
  14     MAX_CACHE_CHARACTER = 10000
  15
  16     def initialize
  17       clear
  18       @parser = CharacterParser.new
  19     end
  20
  21     def clear
  22       @chars = {}
  23     end
  24
  25     def get(s)
  26       check_max
  27       mcs = @parser.parse(s)
  28       @chars[mcs] = Character.new(mcs) if @chars[mcs].nil?
  29       @chars[mcs]
  30     end
  31
  32     def check_max
  33       clear if MAX_CACHE_CHARACTER < @chars.length # clear all cache
  34     end
  35   end
  36
  37   class Character
  38     include UTF8Value
  39     include CharacterIDC
  40     include CharacterIDS
  41
  42     def initialize(char_id)
  43       raise if char_id.nil?
  44       raise unless char_id.kind_of?(Integer) # make sure char_id is Integer.
  45       raise if char_id < 0 # make sure char_id is positive.
  46       @char_id = char_id
  47       @char_id.freeze
  48       # @utf8_mcs = CHISE.i_tou8(@char_id)
  49       @utf8_mcs = itou8(@char_id)
  50       @utf8_mcs.freeze
  51       @feature = {}
  52       # @check_all_done = nil
  53     end
  54     attr_reader :char_id
  55     attr_reader :utf8_mcs
  56
  57     def self.get(s)
  58       CharacterFactory.instance.get(s)
  59     end
  60
  61     def inspect
  62       sprintf("Char:%x", @char_id)
  63     end
  64
  65     def to_s()  @utf8_mcs;      end
  66
  67     def [](f)
  68       f = normalize_feature_name(f)
  69
  70       v = @feature[f]
  71       return v if v
  72       v = @feature["="+f]
  73       return v if v
  74
  75       v = get_feature(f)
  76       if v
  77         @feature[f] = v
  78         return v
  79       end
  80
  81       v = get_feature("="+f)
  82       if v
  83         @feature["="+f] = v
  84         return v
  85       end
  86
  87       nil
  88     end
  89
  90     def []=(k,v)
  91       f = normalize_feature_name(k)
  92       cd = ChiseDB.instance
  93       ft = cd.get_feature(f)
  94       ft.set_value(@char_id, v)
  95       @feature[f] = v;
  96     end
  97
  98     def method_missing(mid, *args) # ref. ostruct.rb
  99       mname = mid.id2name
 100
 101       return self[mname] if args.empty? # get
 102
 103       if args.length == 1 && /=\Z/ =~ mname # put
 104         self[mname.chop] = args.shift
 105         return
 106       end
 107
 108       raise "error"
 109     end
 110
 111     def to_er
 112       en = EntityReferenceEncoder.new
 113       en.to_er(self)
 114     end
 115
 116     def each_feature
 117       cd = ChiseDB.instance
 118       cd.each_feature {|f|
 119         ft = cd.get_feature(f)
 120         begin
 121           v = ft.get_value(@char_id)
 122           next if v.nil?
 123           yield(f, v)
 124         ensure
 125           ft.close
 126         end
 127       }
 128     end
 129
 130     def hash_feature
 131       h = {}
 132       each_feature {|k, v|
 133         h[k] = v
 134       }
 135       h
 136     end
 137
 138     private
 139
 140     def get_feature(f)
 141       cd = ChiseDB.instance
 142       cd.load_feature(f, @char_id)
 143     end
 144
 145     def normalize_feature_name(a)
 146       a = a.gsub(/_/, "-") #underlineは-に置換
 147       a = a.sub(/-at-/,    "@")
 148       a = a.sub(/-plus-/,  "+")
 149       a = a.sub(/\Amap-/,  "=>")
 150       a = a.sub(/\Ato-/,   "->")
 151       a = a.sub(/\Afrom-/, "<-")
 152       a
 153     end
 154
 155   end
 156 end