tagelizer 0.1.3 → 0.1.4

data/VERSION

@@ -1 +1 @@
-0.1.3
+0.1.4

data/lib/tagelizer.rb

@@ -14,7 +14,7 @@ class Tagelizer
   end
 
   def parse( text )
-    text.split(" ").collect {|i| /(\w*)/.match(i)[1]}.select {|i| i.size > @minwordsize}.collect {|w| stemmer.stem corrected_word(w)}
+    remove_duplicates(text.split(" ").collect {|i| /(\w*)/.match(i)[1]}.select {|i| i.size > @minwordsize}.collect {|w| corrected_word(w)})
   end
 
   def speller
@@ -63,6 +63,15 @@ class Tagelizer
     end
   end
 
+  def remove_duplicates list
+    if list.empty?
+      []
+    else
+      tmp = list.pop
+      remove_duplicates(list.select { |word| stemmer.stem(word) != stemmer.stem(tmp) }) + [tmp]
+    end
+  end
+
 
 
 end

data/spec/tagelizer_spec.rb

@@ -22,6 +22,12 @@ describe "Tagelizer" do
   it "can use basic form of words" do
     tagi = Tagelizer.new
     text = "He reads a book."
-    tagi.parse(text).should == ["read", "book"]
+    tagi.parse(text).should == ["reads", "book"]
+  end
+
+  it "should compare stems" do
+    tagi = Tagelizer.new
+    text = "He reads a book as a reading."
+    tagi.parse(text).should == ["book", "reading"]
   end
 end

data/tagelizer.gemspec

@@ -5,7 +5,7 @@
 
 Gem::Specification.new do |s|
   s.name = %q{tagelizer}
-  s.version = "0.1.3"
+  s.version = "0.1.4"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Jonatan Reiners"]

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: tagelizer
 version: !ruby/object:Gem::Version 
-  hash: 29
+  hash: 19
   prerelease: 
   segments: 
   - 0
   - 1
-  - 3
-  version: 0.1.3
+  - 4
+  version: 0.1.4
 platform: ruby
 authors: 
 - Jonatan Reiners