lib/erd_handler/label.rb

   1 class Label
   2
   3   attr_reader :original, :processed
   4
   5   def initialize(original, auto_tidy = false)
   6     @original = original
   7     @processed = [original]
   8     self.tidy if auto_tidy
   9   end
  10
  11   def split(opts = {})
  12     regexp = /[\t\n _,.-]+/
  13     split_camel_case = true
  14     split_numbers = true
  15     if opts.class == Regexp
  16       regexp = opts
  17     else
  18       regexp = opts[:rexexp] if opts.has_key? :regexp
  19       split_camel_case = opts[:camel_case] if opts.has_key? :camel_case
  20       split_numbers = opts[:numbers] if opts.has_key? :numbers
  21     end
  22     @processed = @processed.map do |segment|
  23       segment.split(regexp)
  24     end.flatten
  25
  26     if split_camel_case
  27       @processed = @processed.map do |segment|
  28         segment.split(/(?<=[a-z])(?=[A-Z])/)
  29       end.flatten
  30     end
  31
  32     if split_numbers
  33       @processed = @processed.map do |segment|
  34         segment.split(/(?:(?<!\d)(?=\d))|(?:(?<=\d)(?!\d))/)
  35       end.flatten
  36     end
  37     self
  38   end
  39
  40   def downcase
  41     @processed = @processed.map do |segment| segment.downcase end
  42     self
  43   end
  44
  45   def stem(gb_english = false)
  46     @processed = @processed.map do |segment| segment.stem(gb_english) end
  47     self
  48   end
  49
  50   def tidy
  51     self.split.downcase.stem
  52   end
  53
  54   def length
  55     @processed.join('').length
  56   end
  57
  58   def levenshtein(other_object)
  59     if other_object.class.ancestors.include? Label
  60       other = other_object.processed.join('')
  61     else
  62       other = other_object
  63     end
  64     s = @processed.join('')
  65     n = s.length
  66     m = other.length
  67     return m if (0 == n)
  68     return n if (0 == m)
  69
  70     d = Array.new(m+1) {Array.new(n+1, 0)} # one row for each characer in other, one column for each charater in self
  71
  72     (0..n).each {|i| d[0][i] = i}
  73     (0..m).each {|j| d[j][0] = j}
  74     (1..m).each do |i|
  75       (1..n).each do |j|
  76         d[i][j] = [d[i-1][j-1] + ((s[j-1] == other[i-1]) ? 0 : 1),  # substitution
  77                    d[i-1][j] + 1, # insertion
  78                    d[i][j-1] + 1  # deletion
  79                    ].min
  80       end
  81     end
  82     d[-1][-1]
  83   end
  84
  85   alias :edit_distance :levenshtein
  86 end