lib/wurfl-lite.rb in wurfl-lite-1.0.2 vs lib/wurfl-lite.rb in wurfl-lite-1.1.0

@@ -3,22 +3,49 @@
 require 'hpricot'
 require 'amatch'
 
 class WURFL
 
+  LATEST = 'http://downloads.sourceforge.net/project/wurfl/WURFL/latest/wurfl-latest.xml.gz'
+  BROWSER_PATCH = 'http://wurfl.sourceforge.net/web_browsers_patch.xml'
+
+  attr_accessor :insertion, :substitution, :deletion
+  attr_accessor :match_threshold
+
+  def initialize( filenames = [ LATEST, BROWSER_PATCH ] )
+    clear!
+    [filenames].flatten.each do |filename|
+      process_xml!( filename )
+    end
+    @insertion = 2
+    @substitution = 1
+    @deletion = 1.5
+    @match_threshold = 0.15
+  end
+
   class Hash < ::Hash
     def method_missing( method )
       has_key?( method ) ? self[ method ] : ( self[ :fall_back ] ? self[ :fall_back ].send( method ) : nil )
     end
   end
 
   def clear!
     @devices = Hash.new
     @devices_by_id = {}
+    @files = []
   end
 
+  def reset!
+    files = @files
+    clear!
+    files.each do |filename|
+      process_xml!( filename )
+    end
+  end
+
   def process_xml!( filename )
+    @files << filename
     data = open( filename ).read
     begin # Try decompress it, in case it's a compressed file
       # XXX Yes, this does seem ugly, but is there another way?
       data = Zlib::GzipReader.new(StringIO.new(data.to_s)).read
     rescue Zlib::GzipFile::Error
@@ -53,22 +80,35 @@
       @devices[ key ][ :fall_back ] = @devices_by_id[ @devices[ key ][ :fall_back ] ]
     end
 
   end
 
-  def initialize( filename = 'http://downloads.sourceforge.net/project/wurfl/WURFL/latest/wurfl-latest.xml.gz' )
-    clear!
-    process_xml!( filename )
-    process_xml!( 'http://wurfl.sourceforge.net/web_browsers_patch.xml' )
-  end
-
   def []( user_agent )
     device = @devices[ user_agent ]
-    return device if device
+    if device
+      device[ :wurfl_match ] = { :distance => 0, :distance_normalized => 0 }
+      return device
+    end
     match = Amatch::Sellers.new( user_agent )
+    match.insertion = @insertion
+    match.substitution = @substitution
+    match.deletion = @deletion
     keys = @devices.keys
     distances = match.match( keys )
-    use_key = keys.zip( distances ).sort{|a,b|a.last<=>b.last}.first.first
-    @devices[ user_agent ] = @devices[ use_key ]
+    sorted_list = keys.zip( distances ).sort{|a,b|a.last<=>b.last}
+    use_key = sorted_list.first.first
+    device = @devices[ use_key ]
+    distance = sorted_list.first.last
+    distance_normalised = distance/(user_agent.length+1)
+
+    if distance_normalised > @match_threshold
+      nil
+    else
+      device[ :wurfl_match ] = {
+        :distance => distance,
+        :distance_normalised => distance_normalised
+      }
+      @devices[ user_agent ] = device
+    end
   end
 
 end