# encoding: utf-8
Encoding.default_external = Encoding::UTF_8
file = File.read("x.html").split(/\n/)
c = Hash.new 0
file.each { |line|
  m = line.scan /\p{Cyrillic}+/i
  m.each { |x| c[x] += 1 }
}
 
xwords = ["Каришка", "Липцис", "Андрей", "Воронин", "Пересланные", "сообщения", "что", "меня", "мне", "как"]
 
c.reject {|k, v| xwords.include?(v) || k.length < 3 || v > 30000}.sort_by {|_key, value| value}.last(500).each { |k, v|
  p "#{k} => #{v}"
}

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