wordcut 0.0.4 → 0.0.5

data/wordcut/dict.rb

@@ -1,12 +1,21 @@
 require_relative "dict_seek"
 module Wordcut
   class WordItem
-    attr_reader :headword
+    attr_reader :headword, :payload
     def initialize(headword)
       @headword = headword
-    end
+      @payload = nil
+    end    
   end
 
+  class WordItemWithPayload
+    attr_reader :headword, :payload
+    def initialize(headword, payload)
+      @headword = headword
+      @payload = payload
+    end
+  end
+  
   module DictInfo
     def l
       0
@@ -34,7 +43,9 @@ module Wordcut
       self.concat(open(path).each_line
                    .map(&:strip)
                    .reject(&:empty?)
+                   .sort
                    .map{|w| WordItem.new w})
+
     end
   end
 
@@ -49,4 +60,8 @@ module Wordcut
       return dict
     end
   end
+
+  class DictWithPayload < Array
+    include DictSeeker
+  end
 end

data/wordcut/edge.rb

@@ -3,7 +3,9 @@ module Wordcut
   class Edge
     attr_reader :unk, :chunk, :s, :payload, :etype
     
-    CMP_FUNCS = [lambda {|e| e.unk}, lambda {|e| e.chunk}]
+    CMP_FUNCS = [lambda {|e| e.unk},
+                 lambda {|e| e.chunk},
+                 lambda {|e| e.payload ? 0 : 1}]
     
     def initialize(args = {})
       @unk = args[:unk] || 0

data/wordcut/edge_builder.rb

@@ -13,7 +13,7 @@ module Wordcut
                  :unk => src.unk,
                  :chunk => src.chunk + 1,
                  :etype => :DICT,
-                 :payload => nil)
+                 :payload => pointer.payload)
       end                    
     end
   end

data/wordcut/pointer.rb

@@ -18,6 +18,10 @@ module Wordcut
       final = (@dict[l].headword.length == @offset + 1)
       self.class.new(@s, l, r, @offset + 1, @dict, final)
     end
+
+    def payload
+      @dict[@l].payload
+    end
   end
 
   module PointersManipulator

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: wordcut
 version: !ruby/object:Gem::Version
-  version: 0.0.4
+  version: 0.0.5
 platform: ruby
 authors:
 - Vee Satayamas
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-05-03 00:00:00.000000000 Z
+date: 2016-05-09 00:00:00.000000000 Z
 dependencies: []
 description: Word segmentation tools for ASEAN languages written in Ruby
 email:
@@ -19,6 +19,7 @@ extra_rdoc_files: []
 files:
 - LICENSE
 - README.md
+- data/tha/mixed-tdict.txt
 - data/tha/tdict-acronyms.txt
 - data/tha/tdict-city.txt
 - data/tha/tdict-collection.txt