#!/usr/bin/env ruby
# -*- coding: utf-8 -*-

ru = "/home/pat/repo/translit/corpora/en2ru.txt"

lex = {}

open(ru).map do  |line| 
  source, target = line.split(/\t/)
  lex[source.strip] = target.strip 
end 

def patternize word
 word.split(//).map {|c| word.index(c)}
end

pairs = []

lex.each { |source,target| 
  if patternize(source) == patternize(target)
    source.split(//).zip(target.split(//)).each do |a,b|
      pairs.push [a,b]
    end
  end
}

freq = Hash.new {0}
pairs.each {|pair| freq[pair] += 1 } 

puts freq