lib/spandx/content.rb in spandx-0.4.1 vs lib/spandx/content.rb in spandx-0.5.0

@@ -1,34 +1,41 @@
 # frozen_string_literal: true
 
 module Spandx
   class Content
-    attr_reader :raw, :threshold
+    attr_reader :raw
 
-    def initialize(raw, threshold: 89.0)
-      @threshold = threshold
+    def initialize(raw)
       @raw = raw
     end
 
     def tokens
       @tokens ||= tokenize(canonicalize(raw)).to_set
     end
 
     def similar?(other, algorithm: :dice_coefficient)
       case algorithm
       when :dice_coefficient
-        similarity_score(other) > threshold
+        similarity_score(other, algorithm: algorithm) > 89.0
       when :levenshtein
-        similarity_score(other) < threshold
+        similarity_score(other, algorithm: algorithm) < 3
+      when :jaro_winkler
+        similarity_score(other, algorithm: algorithm) > 89.0
       end
     end
 
     def similarity_score(other, algorithm: :dice_coefficient)
       case algorithm
       when :dice_coefficient
         dice_coefficient(other)
       when :levenshtein
+        require 'text'
+
         Text::Levenshtein.distance(raw, other.raw, 100)
+      when :jaro_winkler
+        require 'jaro_winkler'
+
+        JaroWinkler.distance(raw, other.raw) * 100.0
       end
     end
 
     private