chise/idsdb.rb

   1 # Copyright (C) 2002-2004 Kouichirou Eto, All rights reserved.
   2
   3 require "chise/char"
   4 require "chise/ids"
   5 require "chise/qp"
   6 require "chise/management"
   7
   8 module CHISE
   9   class IDS_DB_Management
  10     def initialize
  11       @cd = ChiseDB.instance
  12       @idsdb = IDS_DB.instance
  13     end
  14
  15     def check_conflict_of_ids_text
  16       @idsdb.each_ccs {|ccs|
  17         qp ccs
  18         c = Hash.new(0)
  19         h = {}
  20         @idsdb.get_ccs(ccs).each_character {|char, ids|
  21           c["char"] += 1
  22           next if ids == char.to_s
  23           next if ids.char_length == 1
  24           char_id = char.char_id
  25           cids = h[char_id]
  26           if cids.nil? # There is no ids yet.
  27             h[char_id] = ids # just set it.
  28             c["good"] += 1
  29           else # but, if there is already a ids?
  30             if cids == ids # the two are same.
  31               c["same"] += 1 # and just ignore
  32             else # but, if the two are not same?
  33               c["conflict"] += 1
  34               puts "conflict\t#{char.to_s}\t#{ids}\t#{cids}"
  35             end
  36           end
  37         }
  38         puts "#{ccs}\t#{c['char']}\t#{c['same']}\t#{c['conflict']}\t#{c['good']}"
  39       }
  40     end
  41
  42     def store_ids_as_text
  43       @idsdb.each_ccs {|ccs|
  44         #qp ccs
  45         @idsdb.get_ccs(ccs).each_character {|char, ids|
  46           next if ids == char.to_s
  47           next if ids.char_length == 1
  48           char.ids_text = ids # just set it.
  49         }
  50       }
  51       @cd.get_feature("ids-text").dump
  52     end
  53
  54     def store_ids_de_er
  55       @cd.get_feature("ids-text").each {|cid, idser|
  56         char = Character.get(cid)
  57         begin
  58           ids = idser.de_er # parse Entity Reference
  59         rescue => e
  60           qp cid, idser
  61           next
  62         end
  63         char.ids_de_er = ids # set it.
  64       }
  65       @cd.get_feature("ids-de-er").dump
  66     end
  67
  68     def check_integrity_of_ids_tree
  69       @cd.get_feature("ids-de-er").each {|cid, ids|
  70         char = Character.get(cid)
  71         idstree = IDS_Tree.new(ids)
  72         begin
  73           raise "contains self" if ids.include?(char.to_s)
  74           idstree.check_integrity
  75         rescue => e
  76           #puts "#{cid}\t#{e.message}\t#{ids}"
  77           char.ids_error = e.message
  78           next
  79         end
  80         char.ids = ids # set it.
  81       }
  82       @cd.get_feature("ids").dump
  83       @cd.get_feature("ids-error").dump
  84     end
  85
  86     def make_by_ids_db
  87       ct = @cd.get_by_ids_db("ids")
  88       @cd.get_feature("ids").each {|cid, ids|
  89         char = Character.get(cid)
  90         ct.set_decoded_char(ids, cid)
  91       }
  92       ct.dump
  93     end
  94
  95     def store_ids_aggregated
  96       @cd.get_feature("ids").each {|cid, ids|
  97         char = Character.get(cid)
  98         #ids = char.decompose
  99         #ids = char.ids
 100         ag = ids.to_ids.aggregate
 101         #puts "#{char.to_s}\t#{ids}\t#{ag}"
 102         char.ids_aggregated = ag
 103       }
 104       @cd.get_feature("ids-aggregated").dump
 105     end
 106
 107     def store_ids_subparts
 108       @cd.get_feature("ids").each {|cid, v|
 109         char = Character.get(cid)
 110         pids = char.to_s # previous_ids
 111         ar = []
 112         i = 0
 113         loop {
 114           ids = pids.decompose
 115           break if ids == pids #これ以上分割できないようだったら終了〜。
 116           ar += ids.to_a
 117           i += 1
 118           qp [char.to_s, pids, ids, ar] if 10 < i #これは何かおかしいぞと
 119           pids = ids
 120         }
 121         str = ar.sort.uniq.join("") # can contain IDC.
 122         char.ids_subparts = str
 123       }
 124       @cd.get_feature("ids-subparts").dump
 125     end
 126
 127     def store_ids_contained
 128       h = Hash.new
 129       @cd.get_feature("ids-subparts").each {|cid, v|
 130         char = Character.get(cid)
 131         parts = char.ids_subparts
 132         parts.each_char {|ch|
 133           h[ch] = [] if h[ch].nil?
 134           h[ch] << cid
 135         }
 136       }
 137       h.each {|ch, v|
 138         #char = Character.get(cid)
 139         char = ch.char
 140         v = v.sort
 141         char.ids_contained = v.join
 142       }
 143       @cd.get_feature("ids-contained").dump
 144     end
 145
 146   end
 147
 148   class IDS_DB
 149     include Singleton
 150
 151     def initialize
 152       @config = Config.instance
 153       @path = @config.ids_dir.path
 154       @dbs = {}
 155     end
 156     attr_reader :path
 157
 158     def get_ccs(ccs)
 159       @dbs[ccs] = IDS_CCS_DB.new(self, ccs) if @dbs[ccs].nil?
 160       @dbs[ccs]
 161     end
 162
 163     def each_ccs
 164       @path.each_entry {|f|
 165         next unless /\AIDS-(.+)\.txt\Z/ =~ f
 166         yield($1)
 167       }
 168     end
 169   end
 170
 171   class IDS_CCS_DB
 172     def initialize(idsdb, ccs)
 173       @idsdb, @ccs = idsdb, ccs
 174       @path = @idsdb.path+("IDS-"+ccs+".txt")
 175     end
 176
 177     def each_line
 178       @path.open {|f|
 179         f.each {|line|
 180           next if /\A;/ =~ line # skip comment
 181           line.chomp!
 182           code, picture, ids = line.split
 183           raise if code.nil?
 184           ids = "" if ids.nil?
 185           yield(code, ids)
 186         }
 187       }
 188     end
 189
 190     def each_character
 191       each_line {|code, ids|
 192         next if ids.nil?
 193         next if ids == "" # If there is no IDS, ignore it.
 194
 195         er = "&"+code+";"
 196         begin
 197           char = Character.get(er)
 198         rescue
 199           #qp er
 200           next
 201         end
 202         next if char.nil?
 203         yield(char, ids)
 204       }
 205     end
 206
 207   end
 208 end