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   end
  95
  96   class IDS_DB
  97     include Singleton
  98
  99     def initialize
 100       @config = Config.instance
 101       @path = @config.ids_dir.path
 102       @dbs = {}
 103     end
 104     attr_reader :path
 105
 106     def get_ccs(ccs)
 107       @dbs[ccs] = IDS_CCS_DB.new(self, ccs) if @dbs[ccs].nil?
 108       @dbs[ccs]
 109     end
 110
 111     def each_ccs
 112       @path.each_entry {|f|
 113         next unless /\AIDS-(.+)\.txt\Z/ =~ f
 114         yield($1)
 115       }
 116     end
 117   end
 118
 119   class IDS_CCS_DB
 120     def initialize(idsdb, ccs)
 121       @idsdb, @ccs = idsdb, ccs
 122       @path = @idsdb.path+("IDS-"+ccs+".txt")
 123     end
 124
 125     def each_line
 126       @path.open {|f|
 127         f.each {|line|
 128           next if /\A;/ =~ line # skip comment
 129           line.chomp!
 130           code, picture, ids = line.split
 131           raise if code.nil?
 132           ids = "" if ids.nil?
 133           yield(code, ids)
 134         }
 135       }
 136     end
 137
 138     def each_character
 139       each_line {|code, ids|
 140         next if ids.nil?
 141         next if ids == "" # If there is no IDS, ignore it.
 142
 143         er = "&"+code+";"
 144         begin
 145           char = Character.get(er)
 146         rescue
 147           #qp er
 148           next
 149         end
 150         next if char.nil?
 151         yield(char, ids)
 152       }
 153     end
 154
 155   end
 156 end