add delete obsolete BDB files function.

diff --git a/src/chise.rb b/src/chise.rb
index 9a95f91..0746c23 100755 (executable)
--- a/src/chise.rb
+++ b/src/chise.rb
@@ -149,7 +149,7 @@ module CHISE
       %w( korean-ksc5601       K0- 4 X),
     ]
     CODESYS_ORDER = %w(japanese chinese korean ideograph)
-    REGEXP_PART = "&([-+0-9A-Za-z]+);"
+    REGEXP_PART = "&([-+0-9A-Za-z#]+);"
     REGEXP_ALL = "^#{REGEXP_PART}$"
 
     def self.match?(er) (er =~ Regexp.new(REGEXP_PART)) != nil end
@@ -158,7 +158,10 @@ module CHISE
       return "" unless er =~ Regexp.new(REGEXP_ALL) #なんか、間違ってる?
       er = $1 #ついでに中身の部分を取り出す
       return $1.hex if er =~ /^MCS-([0-9A-Fa-f]+)/ #MCS
-      return $1.hex if er =~ /^U[-+]?([0-9A-Fa-f]+)/ #Unicode直打ち
+#      if er =~ /^U[-+]?([0-9A-Fa-f]+)/ #Unicode直打ち
+      if er =~ /^U[-+]?([0-9A-Fa-f]+)/ || er =~ /^#([0-9A-Fa-f]+)/ #Unicode直打ち
+       return $1.hex 
+      end
 
       er.sub!(/^I-/, '') if er =~ /^I-/ #I-がついてるとどう違うのかはよくわからない
       each_codesys {|codesys, er_prefix, keta, numtype|        #p [codesys, er_prefix, keta, numtype]
@@ -906,10 +909,8 @@ IDS-CBETA.txt
       print "length    #{h.length}\n"
       cdb = CodesysDB.instance
       cdb.make_db_no_question_mark('ids', h)
-      p ['make db no q mark done.']
       cdb.open_db('ids') #これが無いと、dump_dbされません。
       cdb.dump_db('ids')
-      p ['dump_db done.']
     end
     def char_sort(composed)
       return composed if composed.char_length == 1
@@ -1186,6 +1187,72 @@ IDS-CBETA.txt
     end
   end
 
+  class DBS_MANAGEMENT #======================================================================ファイル管理
+    OBSOLETE_ATTRIBUTES = "
+cns-radical
+cns-radical?
+kangxi-radical
+daikanwa-radical
+unicode-radical
+
+cns-strokes
+kangxi-strokes
+daikanwa-strokes
+shinjigen-1-radical
+gb-original-radical
+japanese-strokes
+jis-strokes-a
+jis-strokes-b
+jisx0208-strokes
+jis-x0213-strokes
+jisx0213-strokes
+unicode-strokes
+
+totalstrokes
+cns-total-strokes
+jis-total-strokes-b
+
+non-morohashi
+
+=>ucs*
+#=>mojikyo
+#=mojikyo
+->identical
+
+ancient-ideograph-of
+ancient-char-of-shinjigen-1
+original-ideograph-of
+original-char-of-shinjigen-1
+simplified-ideograph-of
+vulgar-ideograph-of
+vulgar-char-of-shinjigen-1
+ideograph=
+ideographic-variants
+variant-of-shinjigen-1
+
+iso-10646-comment
+".split
+    def initialize
+      @odir = DB_DIR+"/system-char-id/obsolete" #直打ちしている。
+    end
+    def move_obsolete_files # 廃止予定のbdbファイルをobsoleteディレクトリーにつっこむ
+      db = CharDB.instance
+      db.close_all
+      Dir.mkdir(@odir) unless FileTest.directory? @odir
+      OBSOLETE_ATTRIBUTES.each {|attr|
+       next if attr =~ /^#/
+       filename = db.get_filename(attr)
+       move_to_obsolete(filename)
+       move_to_obsolete(filename+".txt")
+      }
+    end
+    def move_to_obsolete(file)
+      cmd = "mv #{file} #{@odir}"
+#      p cmd
+      system cmd
+    end
+  end
+
 end
 
 #----------------------------------------------------------------------終了

diff --git a/t/tc_char.rb b/t/tc_char.rb
index 364ad2d..8c7e15a 100755 (executable)
--- a/t/tc_char.rb
+++ b/t/tc_char.rb
@@ -103,6 +103,12 @@ class TC_Character < Test::Unit::TestCase
     p_er("&J83-3958;") #購
     p_er("&J90-3958;")
   end
+  def test_read_ucs
+    assert_equal("字", "&U5B57;".de_er)
+    assert_equal("字", "&U-5B57;".de_er)
+    assert_equal("字", "&U+5B57;".de_er)
+    assert_equal("字", "&#5B57;".de_er)
+  end
   def teardown() @char = nil  end
 end
 

diff --git a/tools/make_ids_db.rb b/tools/make_ids_db.rb
index 2508bf2..f515fab 100755 (executable)
--- a/tools/make_ids_db.rb
+++ b/tools/make_ids_db.rb
@@ -8,14 +8,14 @@ require 'chise'
 include CHISE
 
 db = IDS_DB.instance
-db.make_ids_db
-#IDS_TEXT_DB.instance.make_ids_error
-#db.make_ids_reverse
+#db.make_ids_db #1時間12分
+##IDS_TEXT_DB.instance.make_ids_error
+#db.make_ids_reverse #2分
 #db.dump_ids_duplicated #1分
-#db.make_ids_aggregated
+#db.make_ids_aggregated #3分
 #db.dump_ids_aggregated #1分
-#db.make_ids_parts #10分
-#db.make_ids_contained #2分
-#db.make_ids_decomposed #2分
+db.make_ids_parts #10分
+db.make_ids_contained #2分
+db.make_ids_decomposed #2分
 
 #----------------------------------------------------------------------end.