A simple Ruby script to calculate the percentage of similar lines between files. Useful for seeing duplicated lines in rails view files, for example.

similar_lines.rb

# Show the percentage of similar lines between files.
# Requires: diffy gem
# Usage: Edit line 33 to match the files you're interested in, then run this script.

require 'rubygems'
require 'diffy'


def get_dup_lines_percentage(file1, file2)
  starts_with_minus = /^-.*/
  starts_with_plus = /^\+.*/

  diff = Diffy::Diff.new(file1, file2, :source => 'files', context: 0)
  num_lines_in_1 = File.read(file1).lines.count
  num_lines_in_2 = File.read(file2).lines.count

  num_minuses_from_1 = diff.select{|l| starts_with_minus.match(l) }.count
  num_pluses_from_2 = diff.select{|l| starts_with_plus.match(l) }.count

  no_overlap = num_lines_in_1 == num_minuses_from_1

  file1_larger = num_lines_in_1 >= num_lines_in_2

  if file1_larger
    (1 - num_minuses_from_1.to_f / num_lines_in_1 ) * 100
  else
    (1 - num_pluses_from_2.to_f / num_lines_in_2 ) * 100
  end
end


threshold = 50
all_files = Dir.glob '**/*.haml*'

all_files.each do |file_a|
  all_files.each do |file_b|
    next if file_b == file_a
    score = get_dup_lines_percentage(file_a, file_b)
    if score > threshold
      puts "#{file_a} - #{file_b} - #{score}"
    end
  end
end