#!/usr/bin/env ruby

require 'optparse'
require 'ostruct'

configuration = OpenStruct.new

OptionParser.new do |opts|

  opts.banner = "Find similarity in a set of strings."
  opts.separator ''
  opts.separator "Usage: #{File.basename($0)} [OPTIONS]"
  opts.separator ''

  configuration.group = false
  configuration.threshold = 50
  configuration.summary = false

  description = "Group input in batches and process each individually (faster)"
  opts.on("-g", "--group", description) do |v|
    configuration.group = v
  end

  description = "Limit the number of results by threshold, default is 50"
  opts.on("-t", "--threshhold [NUMBER]", OptionParser::DecimalNumeric, description) do |v|
    configuration.threshold = v
  end

  description = "Print a Summary of the groups found"
  opts.on("-s", "--summary", description) do |v|
    configuration.summary = v
  end


  opts.separator ''

end.parse!

# puts configuration.inspect

class Array
  def product
    inject do |cumulative, value|
      cumulative += value
    end
  end
end

class String
  def to_a
    array = []
    size.times do |n|
      array << self[n]
    end
    array
  end

  def scores(other_string)
    longest_string = nil
    if other_string.size > self.size
      longest_string = other_string
      shortest_string = self
    else
      longest_string = self
      shortest_string = other_string
    end
    scores = longest_string.to_a.map do |char|
      0
    end
    shortest_string.size.times do |index|
      if shortest_string[index] == longest_string[index]
        scores[index] = 1
      end
    end
    scores
  end

  def similarity(other_string)
    scores(other_string).product * 100.0 / size
  end
end

class TargetString
  attr_accessor :evaluated
  attr_accessor :data
  def to_s
    data
  end
end

# client

strings = []

STDIN.each_line do |line|
  next if line.chomp == ''
  strings << line.chomp
end

strings.sort! do |n1, n2|
  n1.size <=> n2.size
end
  
strings.reverse!

strings.map! do |string|
  target_string = TargetString.new
  target_string.evaluated = false
  target_string.data = string
  target_string
end

if configuration.group
  groups = strings.group_by do |string|
    string.data.size
  end
else
  groups = { 0 => strings }
end

counter = 0
groups.each do |key, group|
  group.each do |string_1|
    counter = 0
    unless string_1.evaluated
      if configuration.summary
        summary_string = string_1.to_s
      else
        puts "****>>" + string_1.to_s
      end
    end
    string_1.evaluated = true
    group.each do |string_2|
      next if string_2.evaluated
      similarity = string_1.to_s.similarity(string_2.to_s)
      scores = string_1.to_s.scores(string_2.to_s).inspect
      template = "%5.f %s"
      bindings = [similarity, string_2, scores]
      if similarity >= configuration.threshold
        string_2.evaluated = true
        counter += 1
        unless configuration.summary
          puts template % bindings
        end
      end
    end
    if counter > 0
      puts "#{counter} #{summary_string}"
    end
  end
end