sastrawi-ruby 0.2.0

data/lib/sastrawi/dictionary/array_dictionary.rb

@@ -0,0 +1,67 @@
+module Sastrawi
+  module Dictionary
+    class ArrayDictionary
+      attr_reader :words
+
+      def initialize(words = [])
+        @words = []
+
+        add_words(words)
+      end
+
+      ##
+      # Check whether a word is contained in the dictionary
+
+      def contains?(word)
+        @words.include?(word)
+      end
+
+      ##
+      # Count how many words in the dictionary
+
+      def count
+        @words.length
+      end
+
+      ##
+      # Add multiple words to the dictionary
+
+      def add_words(new_words)
+        new_words.each do |word|
+          add(word)
+        end
+      end
+
+      ##
+      # Add a word to the dictionary
+
+      def add(word)
+        return if word.nil? || word.strip == ''
+
+        @words.push(word)
+      end
+
+      ##
+      # Add words from a text file to the dictionary
+
+      def add_words_from_text_file(file_path)
+        words = []
+
+        File.open(file_path, 'r') do |file|
+          file.each do |line|
+            words.push(line.chomp)
+          end
+        end
+
+        add_words(words)
+      end
+
+      ##
+      # Remove a word from the dictionary
+
+      def remove(word)
+        @words.delete(word)
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule10.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule10
+        def disambiguate(word)
+          contains = /^me([lrwy])([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "#{matches[0]}#{matches[1]}#{matches[2]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule11.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule11
+        def disambiguate(word)
+          contains = /^mem([bfv])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule12.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule12
+        def disambiguate(word)
+          contains = /^mempe(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "pe#{matches[0]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule13a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule13a
+        def disambiguate(word)
+          contains = /^mem([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "m#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule13b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule13b
+        def disambiguate(word)
+          contains = /^mem([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "p#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule14.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule14
+        def disambiguate(word)
+          contains = /^men([cdjstz])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule15a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule15a
+        def disambiguate(word)
+          contains = /^men([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "n#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule15b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule15b
+        def disambiguate(word)
+          contains = /^men([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "t#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule16.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule16
+        def disambiguate(word)
+          contains = /^meng([ghqk])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule17a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule17a
+        def disambiguate(word)
+          contains = /^meng([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule17b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule17b
+        def disambiguate(word)
+          contains = /^meng([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "k#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule17c.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule17c
+        def disambiguate(word)
+          contains = /^menge(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return matches[0]
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule17d.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule17d
+        def disambiguate(word)
+          contains = /^meng([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "ng#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule18a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule18a
+        def disambiguate(word)
+          contains = /^meny([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "ny#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule18b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule18b
+        def disambiguate(word)
+          contains = /^meny([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "s#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule19.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule19
+        def disambiguate(word)
+          contains = /^memp([abcdfghijklmopqrstuvwxyz])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "p#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule1a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule1a
+        def disambiguate(word)
+          contains = /^ber([aiueo].*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return matches[0]
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule1b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule1b
+        def disambiguate(word)
+          contains = /^ber([aiueo].*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "r#{matches[0]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule2.rb

@@ -0,0 +1,19 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule2
+        def disambiguate(word)
+          contains = /^ber([bcdfghjklmnpqrstvwxyz])([a-z])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return if /^er(.*)$/.match(matches[2])
+
+            return "#{matches[0]}#{matches[1]}#{matches[2]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule20.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule20
+        def disambiguate(word)
+          contains = /^pe([wy])([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "#{matches[0]}#{matches[1]}#{matches[2]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule21a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule21a
+        def disambiguate(word)
+          contains = /^per([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule21b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule21b
+        def disambiguate(word)
+          contains = /^pe(r[aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule23.rb

@@ -0,0 +1,19 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule23
+        def disambiguate(word)
+          contains = /^per([bcdfghjklmnpqrstvwxyz])([a-z])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return if /^er(.*)$/.match(matches[2])
+
+            return "#{matches[0]}#{matches[1]}#{matches[2]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule24.rb

@@ -0,0 +1,19 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule24
+        def disambiguate(word)
+          contains = /^per([bcdfghjklmnpqrstvwxyz])([a-z])er([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return if matches[0] == 'r'
+
+            return "#{matches[0]}#{matches[1]}er#{matches[2]}#{matches[3]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule25.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule25
+        def disambiguate(word)
+          contains = /^pem([bfv])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule26a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule26a
+        def disambiguate(word)
+          contains = /^pem([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "m#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule26b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule26b
+        def disambiguate(word)
+          contains = /^pem([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "p#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule27.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule27
+        def disambiguate(word)
+          contains = /^pen([cdjstz])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule28a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule28a
+        def disambiguate(word)
+          contains = /^pen([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "n#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule28b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule28b
+        def disambiguate(word)
+          contains = /^pen([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "t#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule29.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule29
+        def disambiguate(word)
+          contains = /^peng([bcdfghjklmnpqrstvwxyz])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule3.rb

@@ -0,0 +1,19 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule3
+        def disambiguate(word)
+          contains = /^ber([bcdfghjklmnpqrstvwxyz])([a-z])er([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return if matches[0] == 'r'
+
+            return "#{matches[0]}#{matches[1]}er#{matches[2]}#{matches[3]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule30a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule30a
+        def disambiguate(word)
+          contains = /^peng([aiueo])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return "#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end