sastrawi-ruby 0.2.0

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule30b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule30b
+        def disambiguate(word)
+          contains = /^peng([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_rule30c.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule30c
+        def disambiguate(word)
+          contains = /^penge(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return matches[0]
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule31a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule31a
+        def disambiguate(word)
+          contains = /^peny([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_rule31b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule31b
+        def disambiguate(word)
+          contains = /^peny([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_rule32.rb

@@ -0,0 +1,19 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule32
+        def disambiguate(word)
+          return 'ajar' if word == 'pelajar'
+
+          contains = /^pe(l[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_rule34.rb

@@ -0,0 +1,19 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule34
+        def disambiguate(word)
+          contains = /^pe([bcdfghjklmnpqrstvwxyz])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return if /^er(.*)$/.match(matches[1])
+
+            return "#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule35.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule35
+        def disambiguate(word)
+          contains = /^ter([bcdfghjkpqstvxz])(er[bcdfghjklmnpqrstvwxyz])(.*)$/.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_rule36.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule36
+        def disambiguate(word)
+          contains = /^pe([bcdfghjkpqstvxz])(er[bcdfghjklmnpqrstvwxyz])(.*)$/.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_rule37a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule37a
+        def disambiguate(word)
+          contains = /^([bcdfghjklmnpqrstvwxyz])(er[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_rule37b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule37b
+        def disambiguate(word)
+          contains = /^([bcdfghjklmnpqrstvwxyz])er([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_rule38a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule38a
+        def disambiguate(word)
+          contains = /^([bcdfghjklmnpqrstvwxyz])(el[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_rule38b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule38b
+        def disambiguate(word)
+          contains = /^([bcdfghjklmnpqrstvwxyz])el([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_rule39a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule39a
+        def disambiguate(word)
+          contains = /^([bcdfghjklmnpqrstvwxyz])(em[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_rule39b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule39b
+        def disambiguate(word)
+          contains = /^([bcdfghjklmnpqrstvwxyz])em([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_rule4.rb

@@ -0,0 +1,11 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule4
+        def disambiguate(word)
+          return 'ajar' if word == 'belajar'
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule40a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule40a
+        def disambiguate(word)
+          contains = /^([bcdfghjklmnpqrstvwxyz])(in[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_rule40b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule40b
+        def disambiguate(word)
+          contains = /^([bcdfghjklmnpqrstvwxyz])in([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_rule41.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule41
+        def disambiguate(word)
+          contains = /^ku(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return matches[0]
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule42.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule42
+        def disambiguate(word)
+          contains = /^kau(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return matches[0]
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule5.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule5
+        def disambiguate(word)
+          contains = /^be([bcdfghjklmnpqstvwxyz])(er[bcdfghjklmnpqrstvwxyz])(.*)$/.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_rule6a.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule6a
+        def disambiguate(word)
+          contains = /^ter([aiueo].*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return matches[0]
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule6b.rb

@@ -0,0 +1,17 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule6b
+        def disambiguate(word)
+          contains = /^ter([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_rule7.rb

@@ -0,0 +1,19 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule7
+        def disambiguate(word)
+          contains = /^ter([bcdfghjklmnpqrstvwxyz])er([aiueo].*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return if matches[0] == 'r'
+
+            return "#{matches[0]}er#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule8.rb

@@ -0,0 +1,19 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule8
+        def disambiguate(word)
+          contains = /^ter([bcdfghjklmnpqrstvwxyz])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return if matches[0] == 'r' || /^er(.*)$/.match(matches[1])
+
+            return "#{matches[0]}#{matches[1]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/disambiguator/disambiguator_prefix_rule9.rb

@@ -0,0 +1,19 @@
+module Sastrawi
+  module Morphology
+    module Disambiguator
+      class DisambiguatorPrefixRule9
+        def disambiguate(word)
+          contains = /^te([bcdfghjklmnpqrstvwxyz])er([bcdfghjklmnpqrstvwxyz])(.*)$/.match(word)
+
+          if contains
+            matches = contains.captures
+
+            return if matches[0] == 'r'
+
+            return "#{matches[0]}er#{matches[1]}#{matches[2]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/morphology/invalid_affix_pair_specification.rb

@@ -0,0 +1,28 @@
+##
+# Asian J. (2007) "Effective Techniques for Indonesian Text Retrieval". page 26
+# http://researchbank.rmit.edu.au/eserv/rmit:6312/Asian.pdf
+
+module Sastrawi
+  module Morphology
+    class InvalidAffixPairSpecification
+      def satisfied_by?(word)
+        return false if /^me(.*)kan$/.match(word)
+
+        return false if word == 'ketahui'
+
+        invalid_affixes = [
+          /^ber(.*)i$/, /^di(.*)an$/, /^ke(.*)i$/, /^ke(.*)an$/,
+          /^me(.*)an$/, /^me(.*)an$/, /^ter(.*)an$/, /^per(.*)an$/
+        ]
+
+        matches = false
+
+        invalid_affixes.each do |invalid_affix|
+          matches = matches || !!(word =~ invalid_affix)
+        end
+
+        matches
+      end
+    end
+  end
+end

data/lib/sastrawi/stemmer/cache/array_cache.rb

@@ -0,0 +1,25 @@
+module Sastrawi
+  module Stemmer
+    module Cache
+      class ArrayCache
+        attr_reader :data
+
+        def initialize
+          @data = {}
+        end
+
+        def set(key, value)
+          @data[key.to_sym] = value
+        end
+
+        def get(key)
+          @data[key.to_sym] if @data.key?(key.to_sym)
+        end
+
+        def has?(key)
+          @data.key?(key.to_sym)
+        end
+      end
+    end
+  end
+end

data/lib/sastrawi/stemmer/cached_stemmer.rb

@@ -0,0 +1,33 @@
+require 'sastrawi/stemmer/filter/text_normalizer'
+
+module Sastrawi
+  module Stemmer
+    class CachedStemmer
+      attr_reader :cache, :delegated_stemmer
+
+      def initialize(cache, delegated_stemmer)
+        @cache = cache
+        @delegated_stemmer = delegated_stemmer
+      end
+
+      def stem(text)
+        normalized_text = Sastrawi::Stemmer::Filter::TextNormalizer.normalize_text(text)
+
+        words = normalized_text.split(' ')
+        stems = []
+
+        words.each do |word|
+          if @cache.has?(word)
+            stems.push(@cache.get(word))
+          else
+            stem = @delegated_stemmer.stem(word)
+            @cache.set(word, stem)
+            stems.push(stem)
+          end
+        end
+
+        stems.join(' ')
+      end
+    end
+  end
+end

data/lib/sastrawi/stemmer/confix_stripping/precedence_adjustment_specification.rb

@@ -0,0 +1,25 @@
+##
+# Confix Stripping Rule Precendence Adjustment Specification
+# Asian J. (2007) "Effective Techniques for Indonesian Text Retrieval" page 78-79
+# http://researchbank.rmit.edu.au/eserv/rmit:6312/Asian.pdf
+
+module Sastrawi
+  module Stemmer
+    module ConfixStripping
+      class PrecedenceAdjustmentSpecification
+        def satisfied_by?(value)
+          regex_rules = [
+            /^be(.*)lah$/, /^be(.*)an$/, /^me(.*)i$/,
+            /^di(.*)i$/, /^pe(.*)i$/, /^ter(.*)i$/
+          ]
+
+          regex_rules.each do |rule|
+            return true if rule.match(value)
+          end
+
+          false
+        end
+      end
+    end
+  end
+end