sweeper 0.2.4 → 0.3

data/CHANGELOG

@@ -1,4 +1,6 @@
 
+v0.3. Pure Ruby Levenshtein algorithm, for Windows support.
+
 v0.2.4. More robust XSD error handling.
 
 v0.2.3. Rescue another OpenURI error.

data/lib/sweeper.rb

@@ -5,7 +5,9 @@ require 'xsd/mapping'
 require 'activesupport'
 require 'open-uri'
 require 'uri'
-require 'amatch'
+require 'Text'
+
+require 'ruby-debug' if ENV['DEBUG']
 
 class ID3Lib::Tag
   def url
@@ -36,8 +38,8 @@ class Sweeper
   def initialize(options = {})
     @dir = File.expand_path(options['dir'] || Dir.pwd)
     @options = options
-    @outf = Tempfile.new("stdout")
-    @errf = Tempfile.new("stderr")    
+    @errf = Tempfile.new("stderr")
+    @match_cache = {}    
   end
   
   # Run the Sweeper according to the <tt>options</tt>.
@@ -94,7 +96,7 @@ class Sweeper
           
         rescue Problem => e          
           tries += 1 and retry if tries < 2
-          puts "Skipped (#{e.message.gsub('\n', ' ')}): #{File.basename(filename)}"
+          puts "Skipped (#{e.message.gsub("\n", " ")}): #{File.basename(filename)}"
           @failed += 1
         end
       end
@@ -194,23 +196,19 @@ class Sweeper
     
     primary = nil
     genres.each_with_index do |this, index|
-      match_results = Amatch::Levenshtein.new(this).similar(GENRES)
-      # Get the levenshtein best-match weight
-      max = match_results.max
-      # Reverse lookup the canonical genre
-      match = GENRES[match_results.index(max)]
+      match, weight = nearest_genre(this)
       # Bias slightly towards higher tagging counts
-      max += ((GENRE_COUNT - index) / GENRE_COUNT / 4.0)
+      weight += ((GENRE_COUNT - index) / GENRE_COUNT / 4.0)
 
       if ['Rock', 'Pop', 'Rap'].include? match
         # Penalize useless genres
-        max = max / 3.0
+        weight = weight / 3.0
       end
             
-      p [max, match] if ENV['DEBUG']
+      p [weight, match] if ENV['DEBUG']
       
-      if !primary or primary.first < max
-        primary = [max, match]
+      if !primary or primary.first < weight
+        primary = [weight, match]
       end
     end
     
@@ -255,4 +253,30 @@ class Sweeper
     ID3Lib::Tag.new(filename, ID3Lib::V_ALL)
   end  
   
+  def nearest_genre(string)
+    @match_cache[string] ||= begin
+      results = {}
+      GENRES.each do |genre|
+        results[Text::Levenshtein.distance(genre, string)] = genre
+      end    
+      min = results.keys.min
+      match = results[min]
+      
+      [match, normalize(match, string, min)]
+    end    
+  end
+  
+  def normalize(genre, string, weight)
+    # XXX Algorithm may not be right
+    if weight == 0
+      1.0
+    elsif weight >= genre.size
+      0.0
+    elsif genre.size >= string.size
+      1.0 - (weight / genre.size.to_f)
+    else
+      1.0 - (weight / string.size.to_f)
+    end    
+  end  
+  
 end

data/sweeper.gemspec

@@ -1,16 +1,16 @@
 
-# Gem::Specification for Sweeper-0.2.4
+# Gem::Specification for Sweeper-0.3
 # Originally generated by Echoe
 
 Gem::Specification.new do |s|
   s.name = %q{sweeper}
-  s.version = "0.2.4"
+  s.version = "0.3"
 
   s.specification_version = 2 if s.respond_to? :specification_version=
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Evan Weaver"]
-  s.date = %q{2008-04-13}
+  s.date = %q{2008-04-16}
   s.default_executable = %q{sweeper}
   s.description = %q{Automatically tag your music collection with metadata from Last.fm.}
   s.email = %q{}
@@ -29,7 +29,7 @@ Gem::Specification.new do |s|
 
   s.add_dependency(%q<id3lib-ruby>, [">= 0"])
   s.add_dependency(%q<choice>, [">= 0"])
-  s.add_dependency(%q<amatch>, [">= 0"])
+  s.add_dependency(%q<Text>, [">= 0"])
   s.add_dependency(%q<activesupport>, [">= 0"])
 end
 
@@ -44,7 +44,7 @@ end
 #   p.summary = "Automatically tag your music collection with metadata from Last.fm."
 #   p.url = "http://blog.evanweaver.com/files/doc/fauna/sweeper/"
 #   p.docs_host = "blog.evanweaver.com:~/www/bax/public/files/doc/"
-#   p.dependencies = ['id3lib-ruby', 'choice', 'amatch', 'activesupport']
+#   p.dependencies = ['id3lib-ruby', 'choice', 'Text', 'activesupport']
 #   p.clean_pattern = ['doc', 'pkg', 'test/integration/songs']
 #   p.rdoc_pattern = ['README', 'LICENSE', 'CHANGELOG', 'TODO', 'lib/*']
 # end

data/test/integration/sweeper_test.rb

@@ -41,6 +41,13 @@ class SweeperTest < Test::Unit::TestCase
     )
   end
   
+  def test_genre_weighting
+    match, weight = @s.nearest_genre('psychedelic')    
+    assert_equal match, 'Psychadelic'
+    assert(weight < 1)
+    assert(weight > 0)
+  end
+  
   def test_read
     assert_equal({}, 
       @s.read(@found_many))

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: sweeper
 version: !ruby/object:Gem::Version 
-  version: 0.2.4
+  version: "0.3"
 platform: ruby
 authors: 
 - Evan Weaver
@@ -30,7 +30,7 @@ cert_chain:
   yZ0=
   -----END CERTIFICATE-----
 
-date: 2008-04-13 00:00:00 -04:00
+date: 2008-04-16 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -52,7 +52,7 @@ dependencies:
         version: "0"
     version: 
 - !ruby/object:Gem::Dependency 
-  name: amatch
+  name: Text
   version_requirement: 
   version_requirements: !ruby/object:Gem::Requirement 
     requirements: