X-Git-Url: https://git.njae.me.uk/?a=blobdiff_plain;f=lib%2Ferd_handler%2Flabel.rb;h=9369190ad12ba4e633e00f74fd0244b6446962b3;hb=HEAD;hp=6c197588c08157320a5fde3c4818db24005a80cc;hpb=55659f547afb1ed903295784f9f4ae99624810d1;p=erd-marker.git

diff --git a/lib/erd_handler/label.rb b/lib/erd_handler/label.rb
index 6c19758..9369190 100644
--- a/lib/erd_handler/label.rb
+++ b/lib/erd_handler/label.rb
@@ -2,27 +2,22 @@ class Label
   
   attr_reader :original, :processed
   
-  def initialize(original)
+  def initialize(original, auto_tidy = false)
     @original = original
     @processed = [original]
+    self.tidy if auto_tidy
   end
   
   def split(opts = {})
+    regexp = /[\t\n _,.-]+/
+    split_camel_case = true
+    split_numbers = true
     if opts.class == Regexp
       regexp = opts
-      split_camel_case = true
     else
-      regexp = opts[:regexp] || /[ _,.-]+/
-      if opts.has_key? :camel_case
-        split_camel_case = opts[:camel_case]
-      else
-        split_camel_case = true
-      end
-      if opts.has_key? :numbers
-        split_numbers = opts[:numbers]
-      else
-        split_numbers = true
-      end
+      regexp = opts[:rexexp] if opts.has_key? :regexp
+      split_camel_case = opts[:camel_case] if opts.has_key? :camel_case
+      split_numbers = opts[:numbers] if opts.has_key? :numbers
     end
     @processed = @processed.map do |segment|
       segment.split(regexp)
@@ -55,4 +50,37 @@ class Label
   def tidy
     self.split.downcase.stem
   end
+  
+  def length
+    @processed.join('').length
+  end
+  
+  def levenshtein(other_object)
+    if other_object.class.ancestors.include? Label
+      other = other_object.processed.join('')
+    else
+      other = other_object
+    end
+    s = @processed.join('')
+    n = s.length
+    m = other.length
+    return m if (0 == n)
+    return n if (0 == m)
+    
+    d = Array.new(m+1) {Array.new(n+1, 0)} # one row for each characer in other, one column for each charater in self
+    
+    (0..n).each {|i| d[0][i] = i}
+    (0..m).each {|j| d[j][0] = j}
+    (1..m).each do |i|
+      (1..n).each do |j|
+        d[i][j] = [d[i-1][j-1] + ((s[j-1] == other[i-1]) ? 0 : 1),  # substitution
+                   d[i-1][j] + 1, # insertion
+                   d[i][j-1] + 1  # deletion
+                   ].min
+      end
+    end
+    d[-1][-1]
+  end
+  
+  alias :edit_distance :levenshtein
 end
\ No newline at end of file