update.

[chise/ruby.git] / chise / ids.rb

diff --git a/chise/ids.rb b/chise/ids.rb
index c2eea75..0ccb8d6 100755 (executable)
--- a/chise/ids.rb
+++ b/chise/ids.rb
@@ -29,7 +29,7 @@ module CHISE
   IDC_SURROUND_FROM_LOWER_LEFT = IDC_A
   IDC_OVERLAID = IDC_B
 
-  class IDS
+  class Nu_IDS
     def initialize(ids)
       @ids = ids
       @ids.freeze
@@ -38,9 +38,11 @@ module CHISE
     def tree() IDS_Tree.new(@ids); end
 
     def compose(dbname="ids")
+      ids = @ids.to_ids.aggregate
+
       cd = ChiseDB.instance
       byidsdb = cd.get_by_ids_db(dbname)
-      cid = byidsdb.decode(@ids)
+      cid = byidsdb.decode(ids)
       return "" if cid.nil? # TO CHECK: why "", not nil?
       composed = Character.get(cid).to_s
       return "" if composed.nil?
@@ -79,6 +81,65 @@ module CHISE
     def decompose_all
       map_char {|ch| ch.char.decompose_all }
     end
+
+    def ids_tree() IDS_Tree.new(self); end
+
+    def compose(dbname="ids")
+      ids = self.aggregate
+      cd = ChiseDB.instance
+      byidsdb = cd.get_by_ids_db(dbname)
+      cid = byidsdb.decode(ids)
+      return "" if cid.nil? # TO CHECK: why "", not nil?
+      composed = Character.get(cid).to_s
+      return "" if composed.nil?
+      return "" if composed.char_length == 0
+      return composed if composed.char_length == 1
+      composed.each_char {|ch|
+       char = ch.char
+       return ch # TO CHECK: the first character?
+      }
+      "" # TO CHECK: why "", not nil?
+    end
+
+    def aggregate(dbname="ids")
+      # In each sub part of IDS, search the corresponding char_id.
+      # If you could search the corresponding char_id, substitute with it.
+      tree = self.ids_tree
+      return self if tree.depth <= 1 # no sub_node
+      tree.sub_nodes.each {|node|
+       c = node.compose(dbname)
+       next if c.nil? || c == ""
+       n = self.gsub(node, c)
+       return n.aggregate(dbname)
+      }
+      self
+    end
+
+    def find() # "日雲"→"曇"とかいう感じの操作
+      ar = []
+      length = char_length()
+      each_char {|ch|
+       char = ch.char
+       ar << char.ids_contained #その文字を含んでいる漢字のリスト
+      }
+      h = Hash.new(0)
+      #qp ar
+      ar.each {|list|
+       next if list.nil?
+       list.each_char {|ch|
+         h[ch] += 1
+       }
+      }
+      str = ""
+      h.each {|k, v|
+       #      p [k, v]
+       if length == v #全部に顔を出していたら
+         str += k
+       end
+      }
+      #    p str
+      str
+    end
   end
 
   module CharacterIDC
@@ -94,12 +155,24 @@ module CHISE
   end
 
   module CharacterIDS
-    def decompose # by glyph
-      decompose_internal
+    def decompose_by_meaning
+      k = self.to_s
+      ids = self.ids_represent
+      return ids if ids && !ids.empty? && k != ids
+      ids = self.ids_element
+      return ids if ids && !ids.empty? && k != ids
+      ids = self.ids_meaning
+      return ids if ids && !ids.empty? && k != ids
+      decompose
     end
 
-    def decompose_by_meaning
-      decompose_internal(true)
+    def decompose # by glyph
+      k = self.to_s
+      ids = self.ids
+      return ids if ids && !ids.empty? && k != ids
+      ids = self.ids_org
+      return ids if ids && !ids.empty? && k != ids
+      k
     end
 
     def decompose_all
@@ -115,43 +188,5 @@ module CHISE
       end
       de
     end
-
-    private
-
-    def decompose_internal(by_meaning=nil)
-      #idss = self.ids
-      #return idss if idss
-      #return k if self.is_basic_kanji?
-      #return ids if idss && 0 < ids.length && k != ids
-
-      k = self.to_s
-      if by_meaning
-       ids = self.ids_represent
-       return ids if ids && 0 < ids.length && k != ids
-       ids = self.ids_element
-       return ids if ids && 0 < ids.length && k != ids
-       ids = self.ids_meaning
-       return ids if ids && 0 < ids.length && k != ids
-      end
-      ids = self.ids
-      return ids if ids && 0 < ids.length && k != ids
-      ids = self.ids_org
-      return ids if ids && 0 < ids.length && k != ids
-      k
-
-      #return k if ids.nil? || ids.length == 0 || k == ids
-      #if ids.char_length == 2
-      #p ["What???", k, ids, k.inspect_all]
-      ##return idsx[1] #二個目だけ返すとか?
-      #return k #IDSに展開する方法が無いと。
-      #end
-      #return k if k == ids
-      #if ids.include?(k) #<C5-4C4D><C6-4A37>この二文字のBUG対策
-      ##return ids.sub(k, "")
-      #return k #IDSに展開する方法が無いと。
-      #end
-      #return ids
-    end
-
   end
 end