spec/groupie_spec.rb in groupie-0.2.2 vs spec/groupie_spec.rb in groupie-0.3.0

@@ -71,13 +71,39 @@
           g[:spam].add %w[word] * 1000
           g[:ham].add %w[word] * 10
           g.classify('word', :log).should == {:spam=>0.75, :ham=>0.25}
         end
       end
+
+      describe "unique" do
+        it "should should behave as sqrt strategy" do
+          g = Groupie.new
+          g[:spam].add %w[buy viagra now]
+          g[:ham].add %w[buy flowers now]
+          g.classify('buy', :unique).should == g.classify('buy', :sqrt)
+          g.classify('flowers', :unique).should == g.classify('flowers', :sqrt)
+        end
+      end
     end
   end
 
+  describe "unique_words" do
+    it "should exclude all words in the 4th quintile of all groups" do
+      g = Groupie.new
+      g[:spam].add %w[one two two three three three four four four four]
+      g[:ham].add %w[apple banana pear orange three]
+      g.unique_words.sort.should == %w[one two apple banana pear orange].sort
+    end
+
+    it "should work on an empty word set" do
+      g = Groupie.new
+      g[:spam].add []
+      g[:ham].add []
+      g.unique_words.should == []
+    end
+  end
+
   context "classify_text" do
     it 'should tokenized html emails' do
       g = Groupie.new
       spam_tokens = File.read(File.join(File.dirname(__FILE__), %w[fixtures spam spam.la-44118014.txt])).tokenize
       ham_tokens = File.read(File.join(File.dirname(__FILE__), %w[fixtures ham spam.la-44116217.txt])).tokenize
@@ -106,25 +132,32 @@
       g = Groupie.new
       g[:spam].add %w[buy viagra now]
       g[:ham].add %w[buy flowers now]
       g.classify_text(%w[buy buckets now]).should == {:spam=>0.5, :ham=>0.5}
     end
-    
+
     it "should support the sqrt strategy" do
       g = Groupie.new
       g[:spam].add %w[one] * 9
       g[:ham].add %w[one]
       g[:spam].add %w[two] * 9
       g[:ham].add %w[two]
       g.classify_text(%w[one two three], :sqrt).should == {:spam=>0.75, :ham=>0.25}
     end
-    
+
     it "should support the log strategy" do
       g = Groupie.new
       g[:spam].add %w[one] * 100
       g[:ham].add %w[one]
       g[:spam].add %w[two]
       g[:ham].add %w[two] * 100
       g.classify_text(%w[one two three], :log).should == {:spam=>0.5, :ham=>0.5}
+    end
+
+    it "should only rate unique words for the unique strategy" do
+      g = Groupie.new
+      g[:spam].add %w[one two two three three three four four four four]
+      g[:ham].add %w[apple banana pear]
+      g.classify_text(%w[one two three apple banana], :unique).should == {:spam=>0.5, :ham=>0.5}
     end
   end
 end